| <desc1 (≤35)> | <desc2 (≤35)> | URL\\n\\nCallouts (≥4, each ≤25 chars):\\n - <callout>\\n\\nRSA1 — [ad group name]\\n Final URL: https://...\\n Path1: ... Path2: ...\\n Headlines (15, each ≤30 chars):\\n 1. <headline> (NN chars)\\n ...\\n 15. <headline> (NN chars)\\n Descriptions (4, each ≤90 chars):\\n 1. <description> (NN chars)\\n ...\\n 4. <description> (NN chars)\\n Pinning: H1=none; H2=none; ... (or explicit pins)\\n\\nRSA2 — ...\\nRSA3 — ...\\n```\\n\\n### Self-check before responding\\n\\nBefore sending the output, run this checklist mentally:\\n\\n- [ ] Each RSA has exactly 15 headlines, exactly 4 descriptions.\\n- [ ] Every headline is ≤30 chars; every description is ≤90 chars. Character counts printed.\\n- [ ] Negative keyword list labeled and ≥8 entries.\\n- [ ] Ad group structure labeled.\\n- [ ] If medical (CFM): no forbidden superlative/outcome words; geo modifier present where required; language is pt-BR.\\n\\nIf any check fails, rewrite before responding. Do not ship partial RSAs.\\n\\n---\\n\\n## Common Mistakes to Avoid\\n\\n### Strategy\\n- Launching without conversion tracking\\n- Too many campaigns (fragmenting budget)\\n- Not giving algorithms enough learning time\\n- Optimizing for wrong metric\\n\\n### Targeting\\n- Audiences too narrow or too broad\\n- Not excluding existing customers\\n- Overlapping audiences competing\\n\\n### Creative\\n- Only one ad per ad set\\n- Not refreshing creative (fatigue)\\n- Mismatch between ad and landing page\\n\\n### Budget\\n- Spreading too thin across campaigns\\n- Making big budget changes (disrupts learning)\\n- Stopping campaigns during learning phase\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What platform(s) are you currently running or want to start with?\\n2. What's your monthly ad budget?\\n3. What does a successful conversion look like (and what's it worth)?\\n4. Do you have existing creative assets or need to create them?\\n5. What landing page will ads point to?\\n6. Do you have pixel/conversion tracking set up?\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md). Key advertising platforms:\\n\\n| Platform | Best For | MCP | Guide |\\n|----------|----------|:---:|-------|\\n| **Google Ads** | Search intent, high-intent traffic | ✓ | [google-ads.md](../../tools/integrations/google-ads.md) |\\n| **Meta Ads** | Demand gen, visual products, B2C | - | [meta-ads.md](../../tools/integrations/meta-ads.md) |\\n| **LinkedIn Ads** | B2B, job title targeting | - | [linkedin-ads.md](../../tools/integrations/linkedin-ads.md) |\\n| **TikTok Ads** | Younger demographics, video | - | [tiktok-ads.md](../../tools/integrations/tiktok-ads.md) |\\n\\nFor tracking setup, see [references/conversion-tracking.md](references/conversion-tracking.md), [ga4.md](../../tools/integrations/ga4.md), [segment.md](../../tools/integrations/segment.md)\\n\\n---\\n\\n## Related Skills\\n\\n- **ad-creative**: For generating and iterating ad headlines, descriptions, and creative at scale\\n- **copywriting**: For landing page copy that converts ad traffic\\n- **analytics**: For proper conversion tracking setup\\n- **ab-testing**: For landing page testing to improve ROAS\\n- **cro**: For optimizing post-click conversion rates\"},\n {\"name\":\"ai-seo\",\"category\":\"SEO\",\"description\":\"\\\"When the user wants to optimize content for AI search engines, get cited by LLMs, or appear in AI-generated answers. Also use when the user mentions 'AI SEO,' 'AEO,' 'GEO,' 'LLMO,' 'answer engine optimization,' 'generative engine optimization,' 'LLM optimization,' 'AI Overviews,' 'optimize for ChatGPT,' 'optimize for Perplexity,' 'AI citations,' 'AI visibility,' 'zero-click search,' 'how do I show up in AI answers,' 'LLM mentions,' or 'optimize for Claude/Gemini.' Use this whenever someone wants their content to be cited or surfaced by AI assistants and AI search engines. For traditional technical and on-page SEO audits, see seo-audit. For structured data implementation, see schema.\\\"\",\"version\":\"2.0.1\",\"body\":\"# AI SEO\\n\\nYou are an expert in AI search optimization — the practice of making content discoverable, extractable, and citable by AI systems including Google AI Overviews, ChatGPT, Perplexity, Claude, Gemini, and Copilot. Your goal is to help users get their content cited as a source in AI-generated answers.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nGather this context (ask if not provided):\\n\\n### 1. Current AI Visibility\\n- Do you know if your brand appears in AI-generated answers today?\\n- Have you checked ChatGPT, Perplexity, or Google AI Overviews for your key queries?\\n- What queries matter most to your business?\\n\\n### 2. Content & Domain\\n- What type of content do you produce? (Blog, docs, comparisons, product pages)\\n- What's your domain authority / traditional SEO strength?\\n- Do you have existing structured data (schema markup)?\\n\\n### 3. Goals\\n- Get cited as a source in AI answers?\\n- Appear in Google AI Overviews for specific queries?\\n- Compete with specific brands already getting cited?\\n- Optimize existing content or create new AI-optimized content?\\n\\n### 4. Competitive Landscape\\n- Who are your top competitors in AI search results?\\n- Are they being cited where you're not?\\n\\n---\\n\\n## How AI Search Works\\n\\n### The AI Search Landscape\\n\\n| Platform | How It Works | Source Selection |\\n|----------|-------------|----------------|\\n| **Google AI Overviews** | Summarizes top-ranking pages | Strong correlation with traditional rankings |\\n| **ChatGPT (with search)** | Searches web, cites sources | Draws from wider range, not just top-ranked |\\n| **Perplexity** | Always cites sources with links | Favors authoritative, recent, well-structured content |\\n| **Gemini** | Google's AI assistant | Pulls from Google index + Knowledge Graph |\\n| **Copilot** | Bing-powered AI search | Bing index + authoritative sources |\\n| **Claude** | Brave Search (when enabled) | Training data + Brave search results |\\n\\nFor a deep dive on how each platform selects sources and what to optimize per platform, see [references/platform-ranking-factors.md](references/platform-ranking-factors.md).\\n\\n### Key Difference from Traditional SEO\\n\\nTraditional SEO gets you ranked. AI SEO gets you **cited**.\\n\\nIn traditional search, you need to rank on page 1. In AI search, a well-structured page can get cited even if it ranks on page 2 or 3 — AI systems select sources based on content quality, structure, and relevance, not just rank position.\\n\\n**Critical stats:**\\n- AI Overviews appear in ~45% of Google searches\\n- AI Overviews reduce clicks to websites by up to 58%\\n- Brands are 6.5x more likely to be cited via third-party sources than their own domains\\n- Optimized content gets cited 3x more often than non-optimized\\n- Statistics and citations boost visibility by 40%+ across queries\\n\\n### Google's Official Stance vs. Multi-Platform Reality\\n\\nThis is important to read once before doing anything else.\\n\\n**Google's position** ([AI features optimization guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)):\\n> \\\"The best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems.\\\"\\n\\nGoogle explicitly says:\\n- **No special markup or files are required** for AI Overviews or AI Mode\\n- **Don't chunk content for AI** — write for people, organize with normal headings and paragraphs\\n- **Don't write separate content for AI** — that risks \\\"scaled content abuse\\\" spam policy\\n- **Helpful, reliable, people-first content** wins — same E-E-A-T standards as regular Search\\n- **No AI-specific Search Console reporting** — use standard SEO metrics\\n\\n**Other AI engines (ChatGPT, Claude, Perplexity, Copilot) behave differently:**\\n- They actively reward extractable structure — passages, FAQs, comparison tables, definition blocks\\n- They parse `llms.txt`, structured pricing pages, and machine-readable files when present\\n- They cite third-party sources (Reddit, Wikipedia, review sites) more heavily than top-ranked pages\\n\\n**What this means for the work:**\\n- The structural patterns in this skill (40–60 word answer blocks, FAQ schema, comparison tables) help **non-Google AI engines** materially. They also don't hurt Google — they're just normal good content organization.\\n- For Google AI Overviews / AI Mode specifically: optimize for people and core Search, full stop. Strong E-E-A-T, original information, semantic HTML, clean indexability.\\n- For ChatGPT/Claude/Perplexity: layer on the extractable structure + llms.txt + machine-readable files.\\n\\nWhen in doubt, default to \\\"write for people, organize for clarity\\\" — that satisfies both camps.\\n\\n### Query Fan-Out (Google AI Search)\\n\\nGoogle's AI features don't just answer the one query a user typed — they generate **concurrent, related queries** under the hood and retrieve results for each.\\n\\nGoogle's own example: a user asking \\\"how to fix lawns\\\" triggers fan-out queries about herbicides, chemical-free removal, weed prevention, etc. The AI synthesizes across all of them.\\n\\n**Implications:**\\n- Single-page-per-keyword targeting is less effective. Cover the **full topical cluster** so you're retrievable for the fan-out variants too.\\n- Long-tail intent matters less than topical authority — Google's AI systems understand synonyms and semantic equivalence.\\n- A page that comprehensively answers a parent topic (with sub-questions covered) will be retrieved more often than narrow per-query pages.\\n\\n**Action**: when planning content, brainstorm the 5–10 related queries the AI is likely to fan out to and make sure your content (or your site as a whole) covers them.\\n\\n---\\n\\n## AI Visibility Audit\\n\\nBefore optimizing, assess your current AI search presence.\\n\\n### Step 1: Check AI Answers for Your Key Queries\\n\\nTest 10-20 of your most important queries across platforms:\\n\\n| Query | Google AI Overview | ChatGPT | Perplexity | You Cited? | Competitors Cited? |\\n|-------|:-----------------:|:-------:|:----------:|:----------:|:-----------------:|\\n| [query 1] | Yes/No | Yes/No | Yes/No | Yes/No | [who] |\\n| [query 2] | Yes/No | Yes/No | Yes/No | Yes/No | [who] |\\n\\n**Query types to test:**\\n- \\\"What is [your product category]?\\\"\\n- \\\"Best [product category] for [use case]\\\"\\n- \\\"[Your brand] vs [competitor]\\\"\\n- \\\"How to [problem your product solves]\\\"\\n- \\\"[Your product category] pricing\\\"\\n\\n### Step 2: Analyze Citation Patterns\\n\\nWhen your competitors get cited and you don't, examine:\\n- **Content structure** — Is their content more extractable?\\n- **Authority signals** — Do they have more citations, stats, expert quotes?\\n- **Freshness** — Is their content more recently updated?\\n- **Schema markup** — Do they have structured data you're missing?\\n- **Third-party presence** — Are they cited via Wikipedia, Reddit, review sites?\\n\\n### Step 3: Content Extractability Check\\n\\nFor each priority page, verify:\\n\\n| Check | Pass/Fail |\\n|-------|-----------|\\n| Clear definition in first paragraph? | |\\n| Self-contained answer blocks (work without surrounding context)? | |\\n| Statistics with sources cited? | |\\n| Comparison tables for \\\"[X] vs [Y]\\\" queries? | |\\n| FAQ section with natural-language questions? | |\\n| Schema markup (FAQ, HowTo, Article, Product)? | |\\n| Expert attribution (author name, credentials)? | |\\n| Recently updated (within 6 months)? | |\\n| Heading structure matches query patterns? | |\\n| AI bots allowed in robots.txt? | |\\n\\n### Step 4: AI Bot Access Check\\n\\nVerify your robots.txt allows AI crawlers. Each AI platform has its own bot, and blocking it means that platform can't cite you:\\n\\n- **GPTBot** and **ChatGPT-User** — OpenAI (ChatGPT)\\n- **PerplexityBot** — Perplexity\\n- **ClaudeBot** and **anthropic-ai** — Anthropic (Claude)\\n- **Google-Extended** — Google Gemini and AI Overviews\\n- **Bingbot** — Microsoft Copilot (via Bing)\\n\\nCheck your robots.txt for `Disallow` rules targeting any of these. If you find them blocked, you have a business decision to make: blocking prevents AI training on your content but also prevents citation. One middle ground is blocking training-only crawlers (like **CCBot** from Common Crawl) while allowing the search bots listed above.\\n\\nSee [references/platform-ranking-factors.md](references/platform-ranking-factors.md) for the full robots.txt configuration.\\n\\n---\\n\\n## Optimization Strategy\\n\\n### The Three Pillars\\n\\n```\\n1. Structure (make it extractable)\\n2. Authority (make it citable)\\n3. Presence (be where AI looks)\\n```\\n\\n### Pillar 1: Structure — Make Content Extractable\\n\\nAI systems extract passages, not pages. Every key claim should work as a standalone statement.\\n\\n**Content block patterns:**\\n- **Definition blocks** for \\\"What is X?\\\" queries\\n- **Step-by-step blocks** for \\\"How to X\\\" queries\\n- **Comparison tables** for \\\"X vs Y\\\" queries\\n- **Pros/cons blocks** for evaluation queries\\n- **FAQ blocks** for common questions\\n- **Statistic blocks** with cited sources\\n\\nFor detailed templates for each block type, see [references/content-patterns.md](references/content-patterns.md).\\n\\n**Structural rules:**\\n- Lead every section with a direct answer (don't bury it)\\n- Keep key answer passages to 40-60 words (optimal for snippet extraction)\\n- Use H2/H3 headings that match how people phrase queries\\n- Tables beat prose for comparison content\\n- Numbered lists beat paragraphs for process content\\n- Each paragraph should convey one clear idea\\n\\n### Pillar 2: Authority — Make Content Citable\\n\\nAI systems prefer sources they can trust. Build citation-worthiness.\\n\\n**The Princeton GEO research** (KDD 2024, studied across Perplexity.ai) ranked 9 optimization methods:\\n\\n| Method | Visibility Boost | How to Apply |\\n|--------|:---------------:|--------------|\\n| **Cite sources** | +40% | Add authoritative references with links |\\n| **Add statistics** | +37% | Include specific numbers with sources |\\n| **Add quotations** | +30% | Expert quotes with name and title |\\n| **Authoritative tone** | +25% | Write with demonstrated expertise |\\n| **Improve clarity** | +20% | Simplify complex concepts |\\n| **Technical terms** | +18% | Use domain-specific terminology |\\n| **Unique vocabulary** | +15% | Increase word diversity |\\n| **Fluency optimization** | +15-30% | Improve readability and flow |\\n| ~~Keyword stuffing~~ | **-10%** | **Actively hurts AI visibility** |\\n\\n**Best combination:** Fluency + Statistics = maximum boost. Low-ranking sites benefit even more — up to 115% visibility increase with citations.\\n\\n**Statistics and data** (+37-40% citation boost)\\n- Include specific numbers with sources\\n- Cite original research, not summaries of research\\n- Add dates to all statistics\\n- Original data beats aggregated data\\n\\n**Expert attribution** (+25-30% citation boost)\\n- Named authors with credentials\\n- Expert quotes with titles and organizations\\n- \\\"According to [Source]\\\" framing for claims\\n- Author bios with relevant expertise\\n\\n**Freshness signals**\\n- \\\"Last updated: [date]\\\" prominently displayed\\n- Regular content refreshes (quarterly minimum for competitive topics)\\n- Current year references and recent statistics\\n- Remove or update outdated information\\n\\n**E-E-A-T alignment**\\n- First-hand experience demonstrated\\n- Specific, detailed information (not generic)\\n- Transparent sourcing and methodology\\n- Clear author expertise for the topic\\n\\n### Pillar 3: Presence — Be Where AI Looks\\n\\nAI systems don't just cite your website — they cite where you appear.\\n\\n**Third-party sources matter more than your own site:**\\n- Wikipedia mentions (7.8% of all ChatGPT citations)\\n- Reddit discussions (1.8% of ChatGPT citations)\\n- Industry publications and guest posts\\n- Review sites (G2, Capterra, TrustRadius for B2B SaaS)\\n- YouTube (frequently cited by Google AI Overviews)\\n- Quora answers\\n\\n**Actions:**\\n- Ensure your Wikipedia page is accurate and current\\n- Participate authentically in Reddit communities\\n- Get featured in industry roundups and comparison articles\\n- Maintain updated profiles on relevant review platforms\\n- Create YouTube content for key how-to queries\\n- Answer relevant Quora questions with depth\\n\\n### Machine-Readable Files for AI Agents\\n\\n> **Google's stance**: not required for AI Overviews or AI Mode. Their guide explicitly says you don't need new markup, AI files, or markdown to appear in generative AI search.\\n>\\n> **Why include them anyway**: non-Google AI engines (ChatGPT, Claude, Perplexity) and autonomous buying agents do reward extractable structure. The files below help with those engines without harming Google.\\n\\nAI agents aren't just answering questions — they're becoming buyers. When an AI agent evaluates tools on behalf of a user, it needs structured, parseable information. If your pricing is locked in a JavaScript-rendered page or a \\\"contact sales\\\" wall, agents will skip you and recommend competitors whose information they can actually read.\\n\\nAdd these machine-readable files to your site root:\\n\\n**`/pricing.md` or `/pricing.txt`** — Structured pricing data for AI agents\\n\\n```markdown\\n# Pricing — [Your Product Name]\\n\\n## Free\\n- Price: $0/month\\n- Limits: 100 emails/month, 1 user\\n- Features: Basic templates, API access\\n\\n## Pro\\n- Price: $29/month (billed annually) | $35/month (billed monthly)\\n- Limits: 10,000 emails/month, 5 users\\n- Features: Custom domains, analytics, priority support\\n\\n## Enterprise\\n- Price: Custom — contact sales@example.com\\n- Limits: Unlimited emails, unlimited users\\n- Features: SSO, SLA, dedicated account manager\\n```\\n\\n**Why this matters now:**\\n- AI agents increasingly compare products programmatically before a human ever visits your site\\n- Opaque pricing gets filtered out of AI-mediated buying journeys\\n- A simple markdown file is trivially parseable by any LLM — no rendering, no JavaScript, no login walls\\n- Same principle as `robots.txt` (for crawlers), `llms.txt` (for AI context), and `AGENTS.md` (for agent capabilities)\\n\\n**Best practices:**\\n- Use consistent units (monthly vs. annual, per-seat vs. flat)\\n- Include specific limits and thresholds, not just feature names\\n- List what's included at each tier, not just what's different\\n- Keep it updated — stale pricing is worse than no file\\n- Link to it from your sitemap and main pricing page\\n\\n**`/llms.txt`** — Context file for AI systems (see [llmstxt.org](https://llmstxt.org))\\n\\nIf you don't have one yet, add an `llms.txt` that gives AI systems a quick overview of what your product does, who it's for, and links to key pages (including your pricing).\\n\\n### Schema Markup for AI\\n\\nStructured data helps AI systems understand your content. Key schemas:\\n\\n| Content Type | Schema | Why It Helps |\\n|-------------|--------|-------------|\\n| Articles/Blog posts | `Article`, `BlogPosting` | Author, date, topic identification |\\n| How-to content | `HowTo` | Step extraction for process queries |\\n| FAQs | `FAQPage` | Direct Q&A extraction |\\n| Products | `Product` | Pricing, features, reviews |\\n| Comparisons | `ItemList` | Structured comparison data |\\n| Reviews | `Review`, `AggregateRating` | Trust signals |\\n| Organization | `Organization` | Entity recognition |\\n\\nContent with proper schema shows 30-40% higher AI visibility on non-Google AI engines. **Google's note**: structured data is \\\"not required for generative AI search\\\" but is recommended for overall SEO strategy. For implementation, use the **schema** skill.\\n\\n---\\n\\n## Agentic Experiences\\n\\nBeyond AI search engines summarizing content, autonomous agents are starting to access sites directly — clicking, reading, comparing, even buying on behalf of users. Google's guide flags this as an emerging category to plan for.\\n\\n**How agents access your site:**\\n- **Visual rendering** — they screenshot/read the page like a user would\\n- **DOM inspection** — they parse the page's HTML structure\\n- **Accessibility tree** — they rely on the same semantic information assistive tech uses (labels, roles, landmarks, headings)\\n\\n**What to do:**\\n- **Render meaningful content without heavy JS gymnastics** — if the page is blank until 4 frameworks finish loading, agents see blank\\n- **Semantic HTML** — use `<main>`, `<nav>`, `<article>`, `<button>`, proper heading hierarchy, `alt` text on images\\n- **Clean accessibility tree** — every interactive element labelled; ARIA used correctly (or not at all when native HTML suffices)\\n- **Stable selectors / predictable layouts** — agents struggle with sites that re-render every interaction\\n- **Visible pricing, specs, contact info** — anything an agent would need to make a buying recommendation should be on a public, indexable page (this is where `/pricing.md` and similar files help)\\n\\n**Emerging — Universal Commerce Protocol (UCP):**\\nGoogle references UCP as a forthcoming protocol that will give agents standardized hooks for commerce interactions (catalog discovery, pricing, checkout). Watch for adoption; for now, the structural recommendations above are the precursor.\\n\\nFor ecom and local business specifically, Google highlights:\\n- **Merchant Center feeds** + **Google Business Profile** for product/service visibility in AI Search\\n- **Business Agent** for conversational customer engagement (where applicable)\\n\\n---\\n\\n## Content Types That Get Cited Most\\n\\nNot all content is equally citable. Prioritize these formats:\\n\\n| Content Type | Citation Share | Why AI Cites It |\\n|-------------|:------------:|----------------|\\n| **Comparison articles** | ~33% | Structured, balanced, high-intent |\\n| **Definitive guides** | ~15% | Comprehensive, authoritative |\\n| **Original research/data** | ~12% | Unique, citable statistics |\\n| **Best-of/listicles** | ~10% | Clear structure, entity-rich |\\n| **Product pages** | ~10% | Specific details AI can extract |\\n| **How-to guides** | ~8% | Step-by-step structure |\\n| **Opinion/analysis** | ~10% | Expert perspective, quotable |\\n\\n**Underperformers for AI citation:**\\n- Generic blog posts without structure\\n- Thin product pages with marketing fluff\\n- Gated content (AI can't access it)\\n- Content without dates or author attribution\\n- PDF-only content (harder for AI to parse)\\n\\n---\\n\\n## Monitoring AI Visibility\\n\\n### What to Track\\n\\n| Metric | What It Measures | How to Check |\\n|--------|-----------------|-------------|\\n| AI Overview presence | Do AI Overviews appear for your queries? | Manual check or Semrush/Ahrefs |\\n| Brand citation rate | How often you're cited in AI answers | AI visibility tools (see below) |\\n| Share of AI voice | Your citations vs. competitors | Peec AI, Otterly, ZipTie |\\n| Citation sentiment | How AI describes your brand | Manual review + monitoring tools |\\n| Source attribution | Which of your pages get cited | Track referral traffic from AI sources |\\n\\n### AI Visibility Monitoring Tools\\n\\n| Tool | Coverage | Best For |\\n|------|----------|----------|\\n| **Otterly AI** | ChatGPT, Perplexity, Google AI Overviews | Share of AI voice tracking |\\n| **Peec AI** | ChatGPT, Gemini, Perplexity, Claude, Copilot+ | Multi-platform monitoring at scale |\\n| **ZipTie** | Google AI Overviews, ChatGPT, Perplexity | Brand mention + sentiment tracking |\\n| **LLMrefs** | ChatGPT, Perplexity, AI Overviews, Gemini | SEO keyword → AI visibility mapping |\\n\\n### DIY Monitoring (No Tools)\\n\\nMonthly manual check:\\n1. Pick your top 20 queries\\n2. Run each through ChatGPT, Perplexity, and Google\\n3. Record: Are you cited? Who is? What page?\\n4. Log in a spreadsheet, track month-over-month\\n\\n### Search Console expectations\\n\\nGoogle's guide is explicit: **there is no AI-specific Search Console reporting**. AI Overviews and AI Mode use core Search ranking, so the standard Search Console reports (Performance, Coverage, Core Web Vitals) are still what you measure with for Google. The third-party tools above are the only way to see cross-platform AI citation behavior.\\n\\n---\\n\\n## What NOT to Do\\n\\nGoogle's guide calls these out explicitly — they hurt across both traditional Search and AI features.\\n\\n1. **Write separate content \\\"for AI\\\"**. Same content should serve people and AI. Writing variants targeted at AI systems risks the **scaled content abuse spam policy** — Google's words.\\n2. **Chunk pages into AI-bait fragments**. Google's guide is direct: *\\\"Don't break your content into tiny pieces for AI to better understand it.\\\"* Use normal paragraph + heading structure.\\n3. **Generate at scale for ranking manipulation**. AI-generated content is fine *if* it meets Search Essentials and spam policies. Mass-producing thin variations does not.\\n4. **Pursue inauthentic mentions**. Don't fabricate citations or bulk-spam Reddit/Wikipedia for AI visibility. Real participation only.\\n5. **Block AI crawlers if you want citation**. Blocking GPTBot, PerplexityBot, ClaudeBot, Google-Extended means those engines literally cannot cite you. Block training-only crawlers (CCBot) if you must, not the search-and-cite ones.\\n6. **Hide your main content behind JS that doesn't render**. Both core Search and AI agents need to see your content; JS-only rendering loses both audiences.\\n7. **Skip E-E-A-T fundamentals**. Author identity, first-hand experience, expertise signals, transparent sourcing — Google's guide leans heavily on these for AI features.\\n\\n---\\n\\n## AI SEO by Content Type\\n\\nFor tactical guidance on SaaS product pages, blog content, comparison/alternative pages, documentation, and local/ecom (Google's emphasis on Merchant Center + Business Profile), see [references/content-types.md](references/content-types.md).\\n\\n---\\n\\n## Common Mistakes\\n\\n- **Ignoring AI search entirely** — ~45% of Google searches now show AI Overviews, and ChatGPT/Perplexity are growing fast\\n- **Treating AI SEO as separate from SEO** — Good traditional SEO is the foundation; AI SEO adds structure and authority on top\\n- **Writing for AI, not humans** — If content reads like it was written to game an algorithm, it won't get cited or convert\\n- **No freshness signals** — Undated content loses to dated content because AI systems weight recency heavily. Show when content was last updated\\n- **Gating all content** — AI can't access gated content. Keep your most authoritative content open\\n- **Ignoring third-party presence** — You may get more AI citations from a Wikipedia mention than from your own blog\\n- **No structured data** — Schema markup gives AI systems structured context about your content\\n- **Keyword stuffing** — Unlike traditional SEO where it's just ineffective, keyword stuffing actively reduces AI visibility by 10% (Princeton GEO study)\\n- **Hiding pricing behind \\\"contact sales\\\" or JS-rendered pages** — AI agents evaluating your product on behalf of buyers can't parse what they can't read. Add a `/pricing.md` file\\n- **Blocking AI bots** — If GPTBot, PerplexityBot, or ClaudeBot are blocked in robots.txt, those platforms can't cite you\\n- **Generic content without data** — \\\"We're the best\\\" won't get cited. \\\"Our customers see 3x improvement in [metric]\\\" will\\n- **Forgetting to monitor** — You can't improve what you don't measure. Check AI visibility monthly at minimum\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md).\\n\\n| Tool | Use For |\\n|------|---------|\\n| `semrush` | AI Overview tracking, keyword research, content gap analysis |\\n| `ahrefs` | Backlink analysis, content explorer, AI Overview data |\\n| `gsc` | Search Console performance data, query tracking |\\n| `ga4` | Referral traffic from AI sources |\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What are your top 10-20 most important queries?\\n2. Have you checked if AI answers exist for those queries today?\\n3. Do you have structured data (schema markup) on your site?\\n4. What content types do you publish? (Blog, docs, comparisons, etc.)\\n5. Are competitors being cited by AI where you're not?\\n6. Do you have a Wikipedia page or presence on review sites?\\n\\n---\\n\\n## Related Skills\\n\\n- **seo-audit**: For traditional technical and on-page SEO audits\\n- **schema**: For implementing structured data that helps AI understand your content\\n- **content-strategy**: For planning what content to create\\n- **competitors**: For building comparison pages that get cited\\n- **programmatic-seo**: For building SEO pages at scale\\n- **copywriting**: For writing content that's both human-readable and AI-extractable\"}\n];\n",[],[352,552,554],"eb52f4d2-5d6d-11f1-95b0-1607ee4eb77e",["D",1780286210534],[444,445],[352,552],["D",1780286210534],[443],[352],"d86ff084-5b4e-11f1-80fc-ee650bb23af1","youtube.ts","d86ff138-5b4e-11f1-80fc-ee650bb23af1",536,["D",1780052962192],"// YouTubeRenderer — per-integration renderer for the 'youtube' integration.\nexport const RENDERER = `\nexport function YouTubeRenderer({data, endpoint}) {\n const YT = '#ff0033';\n const errMsg = data?.error && (typeof data.error === 'string' ? data.error : (data.error.message || JSON.stringify(data.error)));\n if (errMsg) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-start gap-3 border border-error/30'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-on-surface mb-1'}, 'YouTube API error'),\n h('div', {className:'text-[12px] text-text-muted'}, errMsg)\n )\n );\n\n const items = Array.isArray(data?.items) ? data.items : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No results.');\n\n const fmtCount = (v) => {\n const n = parseInt(v, 10);\n if (isNaN(n)) return '-';\n if (n < 1000) return String(n);\n if (n < 1000000) return (n/1000).toFixed(n < 10000 ? 1 : 0) + 'K';\n if (n < 1000000000) return (n/1000000).toFixed(n < 10000000 ? 1 : 0) + 'M';\n return (n/1000000000).toFixed(1) + 'B';\n };\n const fmtDur = (iso) => {\n if (!iso || typeof iso !== 'string') return '';\n const m = iso.match(/PT(?:(\\\\d+)H)?(?:(\\\\d+)M)?(?:(\\\\d+)S)?/);\n if (!m) return '';\n const hh = parseInt(m[1]||'0',10), mm = parseInt(m[2]||'0',10), ss = parseInt(m[3]||'0',10);\n const p2 = (x) => (x < 10 ? '0'+x : ''+x);\n return hh ? (hh + ':' + p2(mm) + ':' + p2(ss)) : (mm + ':' + p2(ss));\n };\n const thumb = (sn) => sn?.thumbnails?.medium?.url || sn?.thumbnails?.high?.url || sn?.thumbnails?.default?.url || '';\n const vidId = (it) => (it.id && typeof it.id === 'object' ? (it.id.videoId || '') : '');\n const chId = (it) => (typeof it.id === 'object' ? (it.id.channelId || '') : (typeof it.id === 'string' ? it.id : ''));\n const chip = (icon, val, color) => h('span', {className:'inline-flex items-center gap-1 text-[10px] font-mono-data text-on-surface-variant'},\n h(Icon, {name:icon, size:11, style:{color: color || YT}}), fmtCount(val)\n );\n\n // --- CHANNELS ----------------------------------------------------------\n if (endpoint === 'channels') {\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(280px, 1fr))'}},\n items.map(ch => {\n const sn = ch.snippet || {}, st = ch.statistics || {};\n const url = sn.customUrl ? 'https://youtube.com/' + sn.customUrl : ('https://youtube.com/channel/' + (typeof ch.id === 'string' ? ch.id : chId(ch)));\n const avatar = thumb(sn);\n return h('a', {key:ch.id?.channelId || ch.id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl p-md flex items-center gap-3 hover:border-primary/40 transition-all group'},\n avatar ? h('img', {src:avatar, className:'w-14 h-14 rounded-full border border-border-subtle flex-shrink-0'})\n : h('div', {className:'w-14 h-14 rounded-full flex items-center justify-center font-bold text-[16px] flex-shrink-0', style:{background:colorFromName(sn.title)+'33', color:colorFromName(sn.title)}}, initialsOf(sn.title)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate flex items-center gap-1'}, sn.title || '(channel)',\n h(Icon, {name:'open_in_new', size:12, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity'})),\n sn.description && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5'}, sn.description),\n h('div', {className:'flex items-center gap-3 mt-1.5'},\n chip('group', st.subscriberCount),\n chip('visibility', st.viewCount, '#94a3b8'),\n st.videoCount != null && chip('movie', st.videoCount, '#94a3b8')\n )\n )\n );\n })\n );\n }\n\n // --- PLAYLISTS ---------------------------------------------------------\n if (endpoint === 'playlists') {\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(300px, 1fr))'}},\n items.map(pl => {\n const sn = pl.snippet || {}, cd = pl.contentDetails || {};\n const id = typeof pl.id === 'string' ? pl.id : '';\n const url = id ? ('https://youtube.com/playlist?list=' + id) : '#';\n const img = thumb(sn);\n return h('a', {key:id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-primary/40 transition-all group'},\n h('div', {className:'relative aspect-video bg-surface-container-high overflow-hidden'},\n img ? h('img', {src:img, className:'w-full h-full object-cover'})\n : h('div', {className:'w-full h-full flex items-center justify-center'}, h(Icon, {name:'playlist_play', size:32, className:'text-text-muted'})),\n cd.itemCount != null && h('div', {className:'absolute right-1.5 bottom-1.5 px-1.5 py-0.5 rounded text-[10px] font-mono-data text-white flex items-center gap-1', style:{background:'rgba(0,0,0,0.78)'}},\n h(Icon, {name:'playlist_play', size:12}), cd.itemCount + ' videos')\n ),\n h('div', {className:'p-md flex flex-col gap-1'},\n h('div', {className:'text-[13px] font-bold text-on-surface leading-snug', style:{display:'-webkit-box', WebkitLineClamp:2, WebkitBoxOrient:'vertical', overflow:'hidden'}}, sn.title || '(untitled playlist)'),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted'},\n h('span', {className:'truncate flex items-center gap-1'}, h(Icon, {name:'account_circle', size:12}), sn.channelTitle || '-'),\n sn.publishedAt && h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(sn.publishedAt))\n )\n )\n );\n })\n );\n }\n\n // --- VIDEOS / SEARCH ---------------------------------------------------\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(300px, 1fr))'}},\n items.map(it => {\n const sn = it.snippet || {}, st = it.statistics || {}, cd = it.contentDetails || {};\n const id = vidId(it) || (typeof it.id === 'string' ? it.id : '');\n const url = id ? ('https://youtube.com/watch?v=' + id) : '#';\n const img = thumb(sn);\n const dur = fmtDur(cd.duration);\n const hasStats = st.viewCount != null || st.likeCount != null || st.commentCount != null;\n return h('a', {key:id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-primary/40 transition-all group'},\n h('div', {className:'relative aspect-video bg-surface-container-high overflow-hidden'},\n img ? h('img', {src:img, className:'w-full h-full object-cover'})\n : h('div', {className:'w-full h-full flex items-center justify-center'}, h(Icon, {name:'movie', size:32, className:'text-text-muted'})),\n dur && h('div', {className:'absolute right-1.5 bottom-1.5 px-1.5 py-0.5 rounded text-[10px] font-mono-data text-white', style:{background:'rgba(0,0,0,0.78)'}}, dur),\n h('div', {className:'absolute inset-0 flex items-center justify-center bg-black/0 group-hover:bg-black/40 transition-colors'},\n h('div', {className:'w-12 h-12 rounded-full flex items-center justify-center opacity-0 group-hover:opacity-100 transition-opacity', style:{background:YT}},\n h(Icon, {name:'play_arrow', size:26, style:{color:'#fff'}}))\n )\n ),\n h('div', {className:'p-md flex flex-col gap-1.5 flex-1'},\n h('div', {className:'text-[13px] font-bold text-on-surface leading-snug', style:{display:'-webkit-box', WebkitLineClamp:2, WebkitBoxOrient:'vertical', overflow:'hidden'}}, sn.title || '(untitled)'),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted'},\n h('span', {className:'truncate flex items-center gap-1'}, h(Icon, {name:'account_circle', size:12}), sn.channelTitle || '-'),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(sn.publishedAt))\n ),\n hasStats && h('div', {className:'flex items-center gap-3 pt-1.5 mt-auto border-t border-border-subtle'},\n st.viewCount != null && chip('visibility', st.viewCount, '#94a3b8'),\n st.likeCount != null && chip('thumb_up', st.likeCount, '#4edea3'),\n st.commentCount != null && chip('forum', st.commentCount, '#c0c1ff')\n )\n )\n );\n })\n );\n}\n`;\n",[],[359,568],"d7f7819e-5b4e-11f1-a8fd-ee650bb23af1","wordpress.ts","d7f78266-5b4e-11f1-a8fd-ee650bb23af1",535,["D",1780052961393],"// WordPressRenderer — per-integration renderer for the 'wordpress' integration.\nexport const RENDERER = `\nexport function WordPressRenderer({data, endpoint}) {\n const WP_BLUE = '#21759b';\n const strip = (html) => (html || '').replace(/<[^>]+>/g, '').replace(/&#?\\\\w+;/g, ' ').replace(/\\\\s+/g, ' ').trim();\n const empty = (label) => h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, label);\n const statusMeta = (s) => s === 'publish' ? {label:'Published', color:'#4edea3'}\n : s === 'draft' ? {label:'Draft', color:'#fbbc04'}\n : s === 'pending' ? {label:'Pending', color:'#8083ff'}\n : s === 'future' ? {label:'Scheduled', color:'#7fc2dc'}\n : s === 'private' ? {label:'Private', color:'#d0bcff'}\n : {label: s || 'Unknown', color:'#94a3b8'};\n\n // Error card\n if (data && !Array.isArray(data) && data.error) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-on-surface'}, 'WordPress error'),\n h('div', {className:'text-[12px] text-text-muted mt-0.5'}, strip(String(data.error)))));\n }\n\n // PUBLISH-POST: single created-post success card\n if (endpoint === 'publish-post' && data && !Array.isArray(data) && data.id) {\n const sm = statusMeta(data.status);\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-emerald-400/40'},\n h('div', {className:'flex items-center gap-2 mb-3'},\n h(Icon, {name:'check_circle', size:18, className:'text-emerald-400'}),\n h('div', {className:'text-[13px] font-bold text-on-surface'}, 'Post created'),\n h('span', {className:'ml-auto inline-flex items-center gap-1.5 px-2.5 py-1 rounded-full text-[10px] font-bold uppercase tracking-wide', style:{background:sm.color+'22', color:sm.color, border:'1px solid '+sm.color+'55'}},\n h('span', {className:'w-1.5 h-1.5 rounded-full', style:{background:sm.color}}), sm.label)),\n h('div', {className:'text-[15px] font-bold text-on-surface'}, strip(data.title?.rendered || data.title?.raw) || '(no title)'),\n strip(data.excerpt?.rendered) && h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 line-clamp-2'}, strip(data.excerpt?.rendered).slice(0,220)),\n h('div', {className:'flex items-center gap-4 mt-3 pt-3 border-t border-border-subtle text-[10px] text-text-muted font-mono-data'},\n h('span', {className:'flex items-center gap-1'}, h(Icon,{name:'calendar_today',size:11}), fmtRelative(data.date) || 'now'),\n h('span', {className:'ml-auto'}, '#' + data.id)),\n data.link && h('a', {href:data.link, target:'_blank', rel:'noreferrer', className:'inline-flex items-center gap-1.5 mt-3 text-[12px] font-medium text-secondary hover:underline'},\n h(Icon,{name:'open_in_new',size:13}), 'View post'));\n }\n\n const rows = Array.isArray(data) ? data : [];\n\n // CATEGORIES: chips with counts\n if (endpoint === 'categories') {\n if (!rows.length) return empty('No categories.');\n const sorted = rows.slice().sort((a, b) => (b.count || 0) - (a.count || 0));\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3'}, sorted.length + ' Categories'),\n h('div', {className:'flex flex-wrap gap-2'},\n sorted.map(c => h('a', {key:c.id, href:c.link || '#', target:'_blank', rel:'noreferrer',\n className:'group inline-flex items-center gap-2 px-3 py-1.5 rounded-full bg-surface-container-high border border-border-subtle hover:border-secondary/50 transition-colors'},\n h('span', {className:'text-[12px] font-medium text-on-surface'}, strip(c.name)),\n h('span', {className:'px-1.5 py-0.5 rounded-full text-[10px] font-mono-data font-bold', style:{background:WP_BLUE+'22', color:'#7fc2dc'}}, (c.count || 0).toLocaleString())))));\n }\n\n // MEDIA: thumbnail grid\n if (endpoint === 'media') {\n if (!rows.length) return empty('No media.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(160px, 1fr))'}},\n rows.map(m => {\n const thumb = m.media_details?.sizes?.medium?.source_url || m.media_details?.sizes?.thumbnail?.source_url || m.source_url;\n const isImg = (m.media_type === 'image') || /^image\\\\//.test(m.mime_type || '');\n const size = m.media_details?.filesize;\n return h('a', {key:m.id, href:m.link || m.source_url || '#', target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-secondary/40 transition-all group'},\n h('div', {className:'aspect-square bg-surface-container flex items-center justify-center overflow-hidden'},\n isImg && thumb\n ? h('img', {src:thumb, loading:'lazy', className:'w-full h-full object-cover group-hover:scale-105 transition-transform', alt:strip(m.title?.rendered)})\n : h(Icon, {name: /^video\\\\//.test(m.mime_type||'') ? 'movie' : /^audio\\\\//.test(m.mime_type||'') ? 'audiotrack' : 'insert_drive_file', size:28, className:'text-text-muted'})),\n h('div', {className:'p-2.5'},\n h('div', {className:'text-[11px] font-bold text-on-surface truncate'}, strip(m.title?.rendered) || 'Untitled'),\n h('div', {className:'flex items-center justify-between text-[9px] text-text-muted font-mono-data mt-1'},\n h('span', {className:'uppercase'}, (m.mime_type || '').split('/')[1] || m.media_type || '-'),\n h('span', null, size ? fmtFileSize(size) : fmtRelative(m.date)))));\n })));\n }\n\n // POSTS / PAGES: post cards\n if (rows.length || endpoint === 'posts' || endpoint === 'pages') {\n if (!rows.length) return empty('No ' + (endpoint || 'posts') + '.');\n return h('div', {className:'mt-3 space-y-3'},\n rows.map(p => {\n const title = strip(p.title?.rendered) || '(no title)';\n const sm = statusMeta(p.status);\n const excerpt = strip(p.excerpt?.rendered);\n return h('div', {key:p.id, className:'glass-panel rounded-xl p-md hover:border-secondary/30 transition-all'},\n h('div', {className:'flex items-start justify-between gap-3'},\n h('a', {href:p.link || '#', target:'_blank', rel:'noreferrer', className:'text-[15px] font-bold text-on-surface hover:text-secondary transition-colors flex items-center gap-1.5 group min-w-0'},\n h('span', {className:'truncate'}, title),\n h(Icon, {name:'open_in_new', size:13, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})),\n h('span', {className:'inline-flex items-center gap-1.5 px-2.5 py-1 rounded-full text-[10px] font-bold uppercase tracking-wide flex-shrink-0', style:{background:sm.color+'22', color:sm.color, border:'1px solid '+sm.color+'55'}},\n h('span', {className:'w-1.5 h-1.5 rounded-full', style:{background:sm.color}}), sm.label)),\n excerpt && h('div', {className:'text-[12px] text-on-surface-variant mt-2 line-clamp-2'}, excerpt.slice(0, 220)),\n h('div', {className:'flex items-center gap-4 mt-2.5 pt-2.5 border-t border-border-subtle text-[10px] text-text-muted font-mono-data'},\n h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'calendar_today', size:11}), fmtRelative(p.date)),\n p.author != null && h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'person', size:11}), 'Author #' + p.author),\n h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'comment', size:11}), (p.comment_count || 0).toLocaleString() + (p.comment_count === 1 ? ' comment' : ' comments')),\n h('span', {className:'ml-auto text-on-surface-variant'}, '#' + p.id)));\n }));\n }\n\n // Generic single-object fallback (default/custom endpoints)\n if (data && !Array.isArray(data) && typeof data === 'object') {\n const keys = Object.keys(data).slice(0, 12);\n if (!keys.length) return empty('No data.');\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3'}, endpoint || 'WordPress'),\n h('div', {className:'space-y-1.5'},\n keys.map(k => {\n const v = data[k];\n const val = (v && typeof v === 'object') ? (v.rendered != null ? strip(v.rendered) : JSON.stringify(v).slice(0,120)) : String(v);\n return h('div', {key:k, className:'flex items-start gap-3 text-[12px] border-b border-border-subtle/50 pb-1.5'},\n h('span', {className:'text-text-muted font-mono-data min-w-[120px]'}, k),\n h('span', {className:'text-on-surface break-all'}, val || '-'));\n })));\n }\n return empty('No data.');\n}\n`;\n",[],[359,576],"d793085e-5b4e-11f1-9305-ee650bb23af1","teams.ts","d793091c-5b4e-11f1-9305-ee650bb23af1",534,["D",1780052960743],"// TeamsRenderer — per-integration renderer for the 'teams' integration.\nexport const RENDERER = `\nexport function TeamsRenderer({data, endpoint}) {\n const {useState} = React;\n const TEAMS_PURPLE = '#6264a7';\n const stripHtml = (s) => (s||'').replace(/<[^>]+>/g,'').replace(/ /g,' ').replace(/&/g,'&').replace(/</g,'<').replace(/>/g,'>').trim();\n\n if (data && data.error) {\n const e = data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[14px]'}, h(Icon,{name:'error',size:18}), 'Teams error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 font-mono-data break-words'}, (e.code ? e.code + ': ' : '') + (e.message || String(e))));\n }\n\n // --- WRITE CONFIRMATIONS ---\n if (endpoint === 'send-channel' || endpoint === 'send-chat') {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-3 border border-secondary/30'},\n h('div', {className:'w-10 h-10 rounded-full flex items-center justify-center flex-shrink-0', style:{background:TEAMS_PURPLE+'22',color:TEAMS_PURPLE}}, h(Icon,{name:'check_circle',size:22})),\n h('div', null,\n h('div', {className:'text-[14px] font-bold text-on-surface'}, 'Message sent'),\n h('div', {className:'text-[11px] text-text-muted mt-0.5'}, data && data.createdDateTime ? fmtRelative(data.createdDateTime) : 'Delivered to Microsoft Teams')));\n }\n\n // --- TEAMS (cards) ---\n if (endpoint === 'teams' || !endpoint) {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No teams joined.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n items.map((t,i) => {\n const name = t.displayName || '(untitled team)';\n return h('div', {key:t.id||i, className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/30 transition-all'},\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center font-bold text-[13px] flex-shrink-0', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, name),\n t.description && h('div', {className:'text-[11px] text-text-muted mt-1 line-clamp-2'}, t.description),\n t.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1.5'}, t.id)));\n })));\n }\n\n // --- CHANNELS (list) ---\n if (endpoint === 'channels') {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No channels.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'divide-y divide-border-subtle/50'},\n items.map((c,i) => h('div', {key:c.id||i, className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors'},\n h('div', {className:'w-8 h-8 rounded-lg flex items-center justify-center flex-shrink-0 mt-0.5', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, h(Icon,{name:'tag',size:16})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, c.displayName || '(unnamed)'),\n c.description && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5'}, c.description),\n c.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1'}, c.id))))));\n }\n\n // --- CHATS (list) ---\n if (endpoint === 'chats') {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No chats.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'divide-y divide-border-subtle/50'},\n items.map((c,i) => {\n const title = c.topic || (c.chatType === 'oneOnOne' ? 'Direct chat' : c.chatType === 'group' ? 'Group chat' : 'Chat');\n return h('div', {key:c.id||i, className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors'},\n h('div', {className:'w-8 h-8 rounded-full flex items-center justify-center flex-shrink-0 mt-0.5', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, h(Icon,{name:c.chatType==='group'?'groups':'chat',size:16})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, title),\n h('div', {className:'flex items-center gap-2 mt-0.5'},\n c.chatType && h('span', {className:'text-[10px] text-text-muted'}, c.chatType),\n c.lastUpdatedDateTime && h('span', {className:'text-[10px] text-text-muted font-mono-data'}, fmtRelative(c.lastUpdatedDateTime))),\n c.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1'}, c.id)));\n })));\n }\n\n // --- MESSAGES + CHAT-MESSAGES (thread + inline compose) ---\n const isChat = endpoint === 'chat-messages';\n const msgs = Array.isArray(data && data.value) ? data.value : [];\n const ctx = msgs[0] || {};\n const ctxTeamId = (ctx.channelIdentity && ctx.channelIdentity.teamId) || '';\n const ctxChannelId = (ctx.channelIdentity && ctx.channelIdentity.channelId) || '';\n const ctxChatId = ctx.chatId || '';\n\n const [teamId, setTeamId] = useState(ctxTeamId);\n const [channelId, setChannelId] = useState(ctxChannelId);\n const [chatId, setChatId] = useState(ctxChatId);\n const [draft, setDraft] = useState('');\n const [sending, setSending] = useState(false);\n const [status, setStatus] = useState('');\n\n const needIds = isChat ? !chatId : (!teamId || !channelId);\n const send = async () => {\n const content = draft.trim();\n if (!content || needIds) return;\n setSending(true); setStatus('');\n try {\n const ep = isChat ? 'send-chat' : 'send-channel';\n const body = isChat ? {chatId, content} : {teamId, channelId, content};\n const r = await fetch('/api/integrations/teams/data?endpoint=' + ep, {\n method:'POST', headers:{'Content-Type':'application/json'}, body: JSON.stringify(body) });\n const j = await r.json().catch(() => ({}));\n if (r.ok && !(j && j.error)) { setStatus('sent'); setDraft(''); }\n else setStatus('Error: ' + ((j && j.error && (j.error.message || j.error)) || ('HTTP ' + r.status)));\n } catch (e) { setStatus('Connection error'); }\n setSending(false);\n };\n\n const idInput = (val, set, ph) => h('input', {value:val, onChange:e=>set(e.target.value), placeholder:ph,\n className:'flex-1 min-w-0 bg-surface-container-lowest border border-border-subtle rounded-lg px-2.5 py-1.5 text-[11px] font-mono-data text-on-surface outline-none focus:border-primary'});\n\n const composer = h('div', {className:'p-3.5 border-t border-border-subtle bg-surface-container-lowest/50 space-y-2'},\n needIds && h('div', {className:'flex gap-2'},\n isChat ? idInput(chatId, setChatId, 'Chat ID')\n : [idInput(teamId, setTeamId, 'Team ID'), idInput(channelId, setChannelId, 'Channel ID')]),\n h('div', {className:'flex items-end gap-2'},\n h('textarea', {value:draft, onChange:e=>setDraft(e.target.value), rows:2,\n placeholder: isChat ? 'Reply to this chat...' : 'Reply to this channel...',\n className:'flex-1 bg-surface-container-lowest border border-border-subtle rounded-lg px-3 py-2 text-[13px] text-on-surface outline-none focus:border-primary resize-none'}),\n h('button', {onClick:send, disabled: sending || !draft.trim() || needIds,\n className:'px-3.5 py-2 rounded-lg text-white text-[13px] font-semibold disabled:opacity-50 flex items-center gap-1.5 flex-shrink-0',\n style:{background:TEAMS_PURPLE}},\n h(Icon,{name:'send',size:15}), sending ? 'Sending...' : 'Send')),\n status && h('div', {className:'text-[11px] ' + (status==='sent' ? 'text-secondary' : 'text-error')},\n status==='sent' ? 'Message sent to Teams.' : status));\n\n if (!msgs.length) {\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'text-center text-text-muted text-[13px] py-8'}, 'No messages yet.'),\n composer);\n }\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex items-center gap-2 bg-surface-container-lowest/50'},\n h(Icon,{name:'forum',size:18,style:{color:'#c0c1ff'}}),\n h('h3', {className:'text-[14px] font-bold'}, msgs.length + ' message' + (msgs.length===1?'':'s'))),\n h('div', {className:'divide-y divide-border-subtle/40 max-h-[480px] overflow-y-auto'},\n msgs.map((m,i) => {\n const sender = (m.from && m.from.user && m.from.user.displayName) || (m.from && m.from.application && m.from.application.displayName) || 'System';\n const text = stripHtml(m.body && m.body.content);\n const isSystem = m.messageType && m.messageType !== 'message';\n const col = colorFromName(sender);\n if (isSystem && !text) return null;\n return h('div', {key:m.id||i, className:'flex items-start gap-3 p-3.5'},\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(sender)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className:'text-[13px] font-bold text-on-surface truncate'}, sender),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.createdDateTime))),\n text ? h('div', {className:'text-[13px] text-on-surface-variant mt-0.5 whitespace-pre-wrap break-words'}, text)\n : h('div', {className:'text-[12px] text-text-muted italic mt-0.5'}, '(no text content)'),\n (m.attachments||[]).length > 0 && h('div', {className:'text-[10px] text-text-muted mt-1 flex items-center gap-1'},\n h(Icon,{name:'attach_file',size:11}), m.attachments.length + ' attachment' + (m.attachments.length===1?'':'s'))));\n })),\n composer);\n}\n`;\n",[],[359,584],"2566a400-5b54-11f1-a375-ee650bb23af1","sheets.ts","d72cca9e-5b4e-11f1-8321-ee650bb23af1",550,["D",1780055238787],"// SheetsRenderer — per-integration renderer for the 'sheets' integration.\nexport const RENDERER = `\nexport function SheetsRenderer({data, endpoint}) {\n if (endpoint === 'values' && Array.isArray(data?.values)) {\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[12px] font-bold'}, data.range),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, data.values.length + ' rows - ' + (data.values[0]?.length || 0) + ' cols')\n ),\n h('div', {className:'overflow-x-auto max-h-[480px]'},\n h('table', {className:'w-full text-[12px]'},\n h('tbody', null,\n data.values.slice(0, 100).map((row, i) => h('tr', {key:i, className:'border-b border-border-subtle/50 hover:bg-white/[0.02]'},\n h('td', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-r border-border-subtle/50', style:{width:40}}, i+1),\n row.map((cell, j) => h('td', {key:j, className:'px-3 py-1.5 ' + (i===0?'font-bold bg-surface-container-lowest':'')}, String(cell)))\n ))\n )\n )\n )\n );\n }\n const sheets = data?.files || [];\n if (!sheets.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No spreadsheets.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(260px, 1fr))'}},\n sheets.map(s => h('a', {\n key:s.id, href:s.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#0f9d5818',color:'#0f9d58'}},\n h(Icon,{name:'table_chart',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, s.name),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted'},\n h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), (s.id||'').slice(0,8) + '-'\n ),\n h('span', null, '-'),\n h('span', null, fmtRelative(s.modifiedTime))\n ),\n s.owners?.[0] && h('div', {className:'text-[10px] text-on-surface-variant truncate mt-1 flex items-center gap-1'},\n h(Icon,{name:'person',size:11}), s.owners[0].displayName\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,592],"d6a99aac-5b4e-11f1-a72a-ee650bb23af1","producthunt.ts","d6a99b74-5b4e-11f1-a72a-ee650bb23af1",532,["D",1780052959207],"// ProductHuntRenderer — per-integration renderer for the 'producthunt' integration.\nexport const RENDERER = `\nexport function ProductHuntRenderer({data, endpoint}) {\n const PH = '#da552f';\n const errs = data?.errors;\n if (Array.isArray(errs) && errs.length) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[13px] mb-2'},\n h(Icon,{name:'error',size:16}), 'Product Hunt API error'),\n h('div', {className:'space-y-1'},\n errs.slice(0,5).map((e,i) => h('div', {key:i, className:'text-[12px] text-on-surface-variant'}, e.message || String(e)))\n )\n );\n }\n\n const topicEdges = data?.data?.topics?.edges;\n if (endpoint === 'topics' || (Array.isArray(topicEdges) && topicEdges.length)) {\n const topics = (topicEdges || []).map(e => e.node || e);\n if (!topics.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No topics.');\n const sorted = [...topics].sort((a,b) => (b.followersCount||0) - (a.followersCount||0));\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-2 mb-3'},\n h(Icon,{name:'tag',size:18,style:{color:PH}}),\n h('h3', {className:'font-bold text-[15px]'}, topics.length + ' Topics')\n ),\n h('div', {className:'flex flex-wrap gap-2'},\n sorted.map(t => h('span', {\n key:t.id || t.slug || t.name,\n className:'inline-flex items-center gap-1.5 px-3 py-1.5 rounded-full text-[12px] font-medium bg-surface-container-high text-on-surface border border-border-subtle'\n },\n t.name,\n t.followersCount != null && h('span', {className:'inline-flex items-center gap-0.5 text-[10px] font-mono-data text-text-muted'},\n h(Icon,{name:'group',size:11}), Number(t.followersCount).toLocaleString()\n )\n ))\n )\n );\n }\n\n const postEdges = data?.data?.posts?.edges || [];\n const posts = postEdges.map(e => e.node || e);\n if (!posts.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No products.');\n\n const chip = (icon, value, color) => h('span', {\n className:'inline-flex items-center gap-1 px-2 py-1 rounded-lg text-[11px] font-bold font-mono-data',\n style:{background:(color||PH)+'18', color:color||PH}\n }, h(Icon,{name:icon,size:12}), Number(value||0).toLocaleString());\n\n return h('div', {className:'mt-3 space-y-2'},\n posts.map((p, idx) => {\n const c = colorFromName(p.name);\n const postTopics = (p.topics?.edges || []).map(e => (e.node||e).name).filter(Boolean).slice(0,3);\n const inner = [\n h('div', {key:'rank', className:'flex flex-col items-center justify-center w-7 flex-shrink-0 text-text-muted font-mono-data text-[13px] font-bold'}, idx + 1),\n p.thumbnail?.url\n ? h('img', {key:'thumb', src:p.thumbnail.url, alt:p.name, className:'w-12 h-12 rounded-lg object-cover flex-shrink-0 border border-border-subtle'})\n : h('div', {key:'thumb', className:'w-12 h-12 rounded-lg flex items-center justify-center font-bold text-[16px] flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(p.name)),\n h('div', {key:'body', className:'flex-1 min-w-0'},\n h('div', {className:'flex items-center gap-2'},\n h('span', {className:'text-[14px] font-bold text-on-surface truncate group-hover:text-secondary transition-colors'}, p.name),\n p.url && h(Icon,{name:'open_in_new',size:13,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n p.tagline && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, p.tagline),\n h('div', {className:'flex items-center gap-1.5 mt-2 flex-wrap'},\n chip('arrow_upward', p.votesCount, PH),\n chip('forum', p.commentsCount, '#94a3b8'),\n postTopics.map(t => h('span', {key:t, className:'px-2 py-0.5 rounded-full text-[10px] text-secondary bg-secondary/10 border border-border-subtle'}, t)),\n p.createdAt && h('span', {className:'text-[10px] text-text-muted font-mono-data ml-auto'}, fmtRelative(p.createdAt))\n )\n )\n ];\n const cls = 'glass-panel rounded-xl p-md flex items-center gap-3 transition-all group ' + (p.url ? 'hover:border-secondary/40 cursor-pointer' : 'border border-border-subtle');\n return p.url\n ? h('a', {key:p.id || idx, href:p.url, target:'_blank', rel:'noreferrer', className:cls}, inner)\n : h('div', {key:p.id || idx, className:cls}, inner);\n })\n );\n}\n`;\n",[],[359,600],"d63bc72a-5b4e-11f1-b9cd-ee650bb23af1","outlook.ts","d63bc806-5b4e-11f1-b9cd-ee650bb23af1",531,["D",1780052958495],"// OutlookRenderer — per-integration renderer for the 'outlook' integration.\nexport const RENDERER = `\nexport function OutlookRenderer({data, endpoint}) {\n const OUTLOOK_BLUE = '#0078d4';\n if (data && data.error) {\n const e = data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[14px]'},\n h(Icon,{name:'error',size:18}), 'Outlook error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 font-mono-data break-words'},\n (e.code ? e.code + ': ' : '') + (e.message || String(e)))\n );\n }\n const stripHtml = (s) => (s||'').replace(/<[^>]+>/g,'').replace(/ /g,' ').replace(/&/g,'&').trim();\n\n // --- CONTACTS --------------------------------------------------------------\n if (endpoint === 'contacts') {\n const items = Array.isArray(data?.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No contacts.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n items.map((c,i) => {\n const name = c.displayName || [c.givenName,c.surname].filter(Boolean).join(' ') || '(no name)';\n const email = c.emailAddresses?.[0]?.address || '';\n const phone = c.mobilePhone || (c.businessPhones||[])[0] || (c.homePhones||[])[0] || '';\n const col = colorFromName(name);\n return h('div', {key:c.id||i, className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/30 transition-all'},\n h('div', {className:'w-11 h-11 rounded-full flex items-center justify-center font-bold text-[13px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, name),\n (c.jobTitle || c.companyName) && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, [c.jobTitle,c.companyName].filter(Boolean).join(' - ')),\n email && h('a', {href:'mailto:'+email, className:'text-[11px] text-secondary truncate mt-1 flex items-center gap-1 hover:underline'},\n h(Icon,{name:'mail',size:11}), email),\n phone && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1 font-mono-data'},\n h(Icon,{name:'call',size:11}), phone)\n )\n );\n })\n )\n );\n }\n\n // --- CALENDAR --------------------------------------------------------------\n if (endpoint === 'calendar') {\n const events = Array.isArray(data?.value) ? data.value : [];\n if (!events.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No events scheduled.');\n const parse = (dt) => dt?.dateTime ? new Date(dt.dateTime + (dt.timeZone === 'UTC' ? 'Z' : '')) : null;\n const groups = {};\n events.forEach(e => {\n const s = parse(e.start);\n const k = s ? s.toDateString() : 'undated';\n (groups[k] = groups[k] || []).push(e);\n });\n const todayStr = new Date().toDateString();\n const tomStr = new Date(Date.now()+86400000).toDateString();\n const dateLabel = (d) => d === todayStr ? 'TODAY' : d === tomStr ? 'TOMORROW' : new Date(d).toLocaleDateString('en-US',{weekday:'long',month:'short',day:'numeric'}).toUpperCase();\n const fmtT = (dt) => { const x = parse(dt); return x ? x.toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : ''; };\n return h('div', {className:'mt-3 space-y-4'},\n Object.entries(groups).map(([k, evs]) => h('div', {key:k},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-secondary mb-2 sticky top-0 bg-background/80 backdrop-blur-sm py-1'},\n dateLabel(k), h('span', {className:'ml-2 text-text-muted'}, '- ' + evs.length + ' event' + (evs.length===1?'':'s'))),\n h('div', {className:'space-y-2'},\n evs.map((e,i) => {\n const att = (e.attendees||[]).filter(a => a.type !== 'resource');\n const join = e.onlineMeeting?.joinUrl;\n return h('a', {key:e.id||i, href:e.webLink || '#', target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-lg p-3 flex items-start gap-3 hover:border-secondary/30 transition-all'},\n h('div', {className:'flex flex-col items-center justify-center w-16 flex-shrink-0 pr-2 border-r border-border-subtle'},\n h('div', {className:'text-[12px] font-bold text-on-surface font-mono-data'}, e.isAllDay ? 'All-day' : fmtT(e.start)),\n !e.isAllDay && fmtT(e.end) && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, '- ' + fmtT(e.end))),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, e.subject || '(no title)'),\n e.location?.displayName && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1'},\n h(Icon,{name:'place',size:11}), e.location.displayName),\n e.organizer?.emailAddress?.name && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, 'Organizer: ' + e.organizer.emailAddress.name),\n h('div', {className:'flex items-center gap-2 mt-1.5 text-[10px] text-text-muted flex-wrap'},\n att.length > 0 && h('span', {className:'flex items-center gap-1'}, h(Icon,{name:'group',size:11}), att.length + ' attendee' + (att.length===1?'':'s')),\n (e.isOnlineMeeting || join) && h('span', {className:'inline-flex items-center gap-0.5 text-secondary'}, h(Icon,{name:'videocam',size:11}), 'Online'))),\n e.showAs && h('div', {className:'w-1.5 h-1.5 rounded-full mt-2 flex-shrink-0', style:{background: e.showAs==='busy' ? '#ffb4ab' : '#4edea3'}}));\n })))\n )\n );\n }\n\n // --- INBOX / MESSAGES (default) -------------------------------------------\n const messages = Array.isArray(data?.value) ? data.value : [];\n if (!messages.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No messages.');\n const unreadCount = messages.filter(m => m.isRead === false).length;\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex justify-between items-center bg-surface-container-lowest/50'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'},\n h(Icon,{name:'inbox',size:18, style:{color:OUTLOOK_BLUE}}),\n messages.length.toLocaleString() + ' messages'),\n unreadCount > 0 && h('span', {className:'text-[10px] font-mono-data px-2 py-0.5 rounded-full', style:{background:OUTLOOK_BLUE+'22',color:OUTLOOK_BLUE}}, unreadCount + ' unread')),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n messages.map((m,i) => {\n const ea = m.from?.emailAddress || m.sender?.emailAddress || {};\n const fromName = ea.name || ea.address || '(unknown)';\n const unread = m.isRead === false;\n const col = colorFromName(fromName);\n const preview = stripHtml(m.bodyPreview);\n return h('a', {key:m.id||i, href:m.webLink || '#', target:'_blank', rel:'noreferrer',\n className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors ' + (unread ? 'border-l-2' : ''),\n style: unread ? {borderLeftColor:OUTLOOK_BLUE} : null},\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(fromName)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className: 'text-[13px] truncate ' + (unread?'font-bold text-on-surface':'text-on-surface-variant')}, fromName),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.receivedDateTime))),\n h('div', {className:'text-[13px] truncate ' + (unread?'font-semibold text-on-surface':'text-on-surface-variant')}, m.subject || '(no subject)'),\n preview && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, preview),\n (m.hasAttachments || m.importance === 'high' || m.flag?.flagStatus === 'flagged') && h('div', {className:'flex flex-wrap gap-1 mt-1.5'},\n m.importance === 'high' && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-error bg-error/10 border border-error/30'}, h(Icon,{name:'priority_high',size:10}), 'HIGH'),\n m.hasAttachments && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-on-surface-variant bg-surface-container-high'}, h(Icon,{name:'attach_file',size:10}), 'ATTACHMENT'),\n m.flag?.flagStatus === 'flagged' && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-tertiary bg-tertiary/10 border border-tertiary/30'}, h(Icon,{name:'flag',size:10}), 'FLAGGED'))\n )\n );\n })\n )\n );\n}\n`;\n",[],[359,608],"d5d5ef7c-5b4e-11f1-95e0-ee650bb23af1","onedrive.ts","d5d5f026-5b4e-11f1-95e0-ee650bb23af1",530,["D",1780052957825],"// OneDriveRenderer — per-integration renderer for the 'onedrive' integration.\nexport const RENDERER = `\nexport function OneDriveRenderer({data, endpoint}) {\n if (data && data.error) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 text-[13px] text-error'}, 'OneDrive error: ' + (data.error.message || data.error.code || String(data.error)));\n // createFolder returns the new driveItem (single object) → confirmation card.\n if (endpoint === 'createFolder' && data && data.id && data.name) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-4'},\n h('div', {className:'w-12 h-12 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#0078d418',color:'#0078d4'}}, h(Icon,{name:'create_new_folder',size:24})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, 'Folder created: ' + data.name),\n data.webUrl && h('a', {href:data.webUrl, target:'_blank', rel:'noopener', className:'text-[11px] text-primary truncate inline-flex items-center gap-1 mt-0.5'}, h(Icon,{name:'open_in_new',size:11}), 'Open in OneDrive')\n )\n );\n }\n const items = (data && data.value) || [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No files.');\n const mimeIcon = (mime) => {\n if (!mime) return {icon:'insert_drive_file', color:'#94a3b8'};\n if (mime.indexOf('pdf') !== -1) return {icon:'picture_as_pdf', color:'#ea4335'};\n if (mime.indexOf('word') !== -1 || mime.indexOf('document') !== -1) return {icon:'description', color:'#2b579a'};\n if (mime.indexOf('sheet') !== -1 || mime.indexOf('excel') !== -1 || mime.indexOf('csv') !== -1) return {icon:'table_chart', color:'#217346'};\n if (mime.indexOf('presentation') !== -1 || mime.indexOf('powerpoint') !== -1) return {icon:'co_present', color:'#d24726'};\n if (mime.indexOf('zip') !== -1 || mime.indexOf('compressed') !== -1) return {icon:'folder_zip', color:'#5f6368'};\n if (mime.indexOf('image/') === 0) return {icon:'image', color:'#34a853'};\n if (mime.indexOf('video/') === 0) return {icon:'movie', color:'#ea4335'};\n if (mime.indexOf('audio/') === 0) return {icon:'audiotrack', color:'#a078ff'};\n return {icon:'insert_drive_file', color:'#94a3b8'};\n };\n return h('div', {className:'mt-3'},\n h('div', {className:'flex items-center justify-between mb-3'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'}, h(Icon,{name:'cloud',size:18,className:'text-[#0078d4]'}), items.length + (items.length === 1 ? ' item' : ' items')),\n h('span', {className:'text-[10px] font-mono-data text-text-muted uppercase tracking-widest'}, endpoint || 'recent')\n ),\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(220px, 1fr))'}},\n items.map(f => {\n const isFolder = !!f.folder;\n const ico = isFolder ? {icon:'folder', color:'#ffc83d'} : mimeIcon(f.file && f.file.mimeType);\n const ownerName = (f.createdBy && f.createdBy.user && f.createdBy.user.displayName) || (f.lastModifiedBy && f.lastModifiedBy.user && f.lastModifiedBy.user.displayName) || '-';\n const c = colorFromName(ownerName);\n const subtitle = isFolder ? ((f.folder.childCount != null ? f.folder.childCount + ' items' : 'Folder')) : ((f.file && f.file.mimeType) ? f.file.mimeType.replace(/^application\\\\//,'').replace(/^vnd\\\\.openxmlformats-officedocument\\\\./,'') : 'File');\n return h('a', {\n key: f.id, href: f.webUrl || '#', target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex flex-col gap-2 hover:border-primary/30 transition-all group'\n },\n h('div', {className:'flex items-start justify-between gap-2'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background: ico.color+'18', color: ico.color}}, h(Icon,{name: ico.icon, size:22})),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, f.name),\n h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, subtitle)\n ),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted pt-2 border-t border-border-subtle'},\n h('span', {className:'truncate flex items-center gap-1'},\n h('span', {className:'w-4 h-4 rounded-full flex items-center justify-center text-[7px] font-bold flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(ownerName)),\n ownerName\n ),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(f.lastModifiedDateTime))\n ),\n (!isFolder && f.size) ? h('div', {className:'flex justify-end'}, h('span', {className:'font-mono-data text-[9px] text-on-surface-variant'}, fmtFileSize(f.size))) : (isFolder ? h('div', {className:'flex justify-end'}, h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-[#0078d4]/10 text-[#0078d4]'}, h(Icon,{name:'folder',size:9}), 'FOLDER')) : null)\n );\n })\n )\n );\n}\n`;\n",[],[359,616],"d57a019e-5b4e-11f1-89bf-ee650bb23af1","linkedin.ts","d57a0248-5b4e-11f1-89bf-ee650bb23af1",529,["D",1780052957225],"// LinkedInRenderer — per-integration renderer for the 'linkedin' integration.\nexport const RENDERER = `\nexport function LinkedInRenderer({data, endpoint}) {\n const {useState} = React;\n const LI_BLUE = '#0077b5';\n const LI_LIGHT = '#00a0dc';\n\n // Local tab: profile | posts | analytics | compose\n const [tab, setTab] = useState(endpoint === 'analytics' ? 'analytics' : endpoint === 'posts' ? 'posts' : 'profile');\n const [draft, setDraft] = useState('');\n const [posting, setPosting] = useState(false);\n const [posted, setPosted] = useState('');\n\n // -- helpers --------------------------------------------------------------\n const fmtDate = (ms) => {\n if (!ms) return '-';\n try { return new Date(Number(ms)).toLocaleDateString(undefined, {month:'short', day:'numeric', year:'numeric'}); }\n catch { return '-'; }\n };\n\n const lifecycleBadge = (state) => {\n const color = state === 'PUBLISHED' ? '#4edea3' : state === 'DRAFT' ? '#fbbc04' : '#94a3b8';\n return h('span', {style:{background:color+'22',color,border:'1px solid '+color+'66',borderRadius:4,padding:'1px 7px',fontSize:10,fontWeight:700,letterSpacing:.5}}, state || '-');\n };\n\n // -- tab bar --------------------------------------------------------------\n const tabBtn = (id, label) => h('button', {\n key:id,\n onClick:() => setTab(id),\n style:{\n padding:'5px 16px', borderRadius:20, fontSize:12, fontWeight:700, cursor:'pointer',\n background: tab===id ? LI_BLUE : 'transparent',\n color: tab===id ? '#fff' : LI_BLUE,\n border: '1px solid '+LI_BLUE,\n transition:'all .15s'\n }\n }, label);\n\n const tabRow = h('div', {style:{display:'flex',gap:8,marginBottom:16,flexWrap:'wrap'}},\n tabBtn('profile','Profile'),\n tabBtn('posts','Posts'),\n tabBtn('analytics','Analytics'),\n tabBtn('compose','Compose')\n );\n\n // -- compose view: draft a post and queue it for approval -----------------\n const composeView = () => {\n const submit = async () => {\n const text = draft.trim();\n if (!text) return;\n setPosting(true); setPosted('');\n try {\n const r = await fetch('/api/actions', {\n method:'POST', headers:{'Content-Type':'application/json'},\n body: JSON.stringify({ integration:'linkedin', endpoint:'create-post', source_type:'linkedin', workspace:'sales', title:'LinkedIn post', preview: text.slice(0,140), payload:{ text } })\n });\n if (r.ok) { setPosted('Queued in Approvals for one-click publish.'); setDraft(''); }\n else { const j = await r.json().catch(() => ({})); setPosted('Error: ' + (j.error || ('HTTP ' + r.status))); }\n } catch (e) { setPosted('Connection error'); }\n setPosting(false);\n };\n return h('div', {className:'mt-3'},\n h('div', {className:'glass-panel rounded-xl p-lg space-y-3'},\n h('div', {className:'flex items-center gap-2'},\n h('div', {className:'w-8 h-8 rounded-full flex items-center justify-center text-[12px] font-bold', style:{background:'linear-gradient(135deg,'+LI_BLUE+','+LI_LIGHT+')', color:'#fff'}}, 'in'),\n h('span', {className:'text-[13px] font-bold text-on-surface'}, 'Compose LinkedIn post')\n ),\n h('textarea', {\n value: draft, onChange: e => setDraft(e.target.value), rows:6, maxLength:3000,\n placeholder:'Share an update on LinkedIn...',\n className:'w-full bg-surface-container-lowest border border-border-subtle rounded-lg px-3 py-2.5 text-[13px] text-on-surface outline-none focus:border-primary resize-none'\n }),\n h('div', {className:'flex items-center justify-between gap-3'},\n h('span', {className:'text-[11px] text-text-muted font-mono-data'}, draft.length + ' / 3000'),\n h('button', {\n onClick: submit, disabled: posting || !draft.trim(),\n className:'px-4 py-2 rounded-lg text-white text-[13px] font-semibold disabled:opacity-50',\n style:{background:LI_BLUE}\n }, posting ? 'Queuing...' : 'Queue post')\n ),\n posted && h('div', {className:'text-[12px] ' + (posted.startsWith('Queued') ? 'text-secondary' : 'text-error')}, posted),\n h('div', {className:'text-[11px] text-text-muted'}, 'Nothing is posted to LinkedIn until you approve it in Approvals.')\n )\n );\n };\n\n // -- profile view ---------------------------------------------------------\n const profileView = () => {\n const name = data?.name || [data?.localizedFirstName, data?.localizedLastName].filter(Boolean).join(' ') || '-';\n const head = data?.headline || data?.localizedHeadline || '';\n const lid = data?.id || '';\n const c = colorFromName(name);\n return h('div', {className:'glass-panel rounded-xl p-lg flex items-center gap-4 mt-3'},\n h('div', {\n className:'w-14 h-14 rounded-full flex items-center justify-center font-bold text-[18px] flex-shrink-0',\n style:{background:'linear-gradient(135deg,'+LI_BLUE+','+LI_LIGHT+')', color:'#fff'}\n }, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('h3', {className:'font-bold text-[18px] text-on-surface'}, name),\n head && h('div', {className:'text-[13px] text-on-surface-variant mt-0.5'}, head),\n lid && h('div', {className:'text-[11px] text-text-muted font-mono-data mt-1'}, 'ID: ' + lid)\n )\n );\n };\n\n // -- posts view -----------------------------------------------------------\n const postsView = () => {\n const items = Array.isArray(data?.elements) ? data.elements : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No posts found.');\n return h('div', {className:'mt-3 space-y-2 max-h-[520px] overflow-y-auto'},\n items.map((p,i) => {\n const urn = (p.id || p.urn || '').toString();\n const ts = p.created?.time;\n const state = p.lifecycleState || p.visibility || '';\n return h('div', {key:i, className:'glass-panel rounded-xl p-md border border-border-subtle'},\n h('div', {className:'flex items-center justify-between gap-3 flex-wrap'},\n h('span', {className:'text-[12px] text-text-muted font-mono-data'}, fmtDate(ts)),\n lifecycleBadge(state)\n ),\n urn && h('div', {className:'flex items-center justify-between gap-2 mt-1.5'},\n h('span', {className:'text-[11px] text-text-muted font-mono-data truncate'}, urn.length > 50 ? urn.slice(0,50)+'...' : urn),\n h('a', {href:'https://www.linkedin.com/feed/update/'+encodeURIComponent(urn), target:'_blank', rel:'noreferrer', className:'text-[11px] font-semibold flex items-center gap-1 flex-shrink-0', style:{color:LI_BLUE}}, h(Icon,{name:'open_in_new',size:12}), 'Open')\n )\n );\n })\n );\n };\n\n // -- analytics view -------------------------------------------------------\n const analyticsView = () => {\n const items = Array.isArray(data?.elements) ? data.elements : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No analytics data.');\n const totals = items.reduce((acc, el) => {\n const s = el.shareStatistics || {};\n acc.impressions += (s.impressionCount || 0);\n acc.clicks += (s.clickCount || 0);\n acc.likes += (s.likeCount || 0);\n acc.comments += (s.commentCount || 0);\n return acc;\n }, {impressions:0,clicks:0,likes:0,comments:0});\n const chip = (label, value, icon) => h('div', {className:'glass-panel rounded-xl p-md flex flex-col items-center gap-1 border border-border-subtle min-w-[110px]'},\n h('span', {className:'text-[24px] font-bold text-on-surface'}, value.toLocaleString()),\n h('span', {className:'text-[11px] text-text-muted uppercase tracking-widest'}, label)\n );\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(110px, 1fr))'}},\n chip('Impressions', totals.impressions),\n chip('Clicks', totals.clicks),\n chip('Likes', totals.likes),\n chip('Comments', totals.comments)\n ),\n h('div', {className:'text-[11px] text-text-muted mt-3 text-right'}, items.length + ' post(s) aggregated')\n );\n };\n\n return h('div', {className:'text-on-surface'},\n tabRow,\n tab === 'profile' ? profileView() :\n tab === 'posts' ? postsView() :\n tab === 'compose' ? composeView() :\n analyticsView()\n );\n}\n`;\n",[],[359,624],"d50ecbe0-5b4e-11f1-89bf-ee650bb23af1","gsc.ts","d50eccb2-5b4e-11f1-89bf-ee650bb23af1",528,["D",1780052956519],"// GSCRenderer — per-integration renderer for the 'gsc' integration.\nexport const RENDERER = `\nexport function GSCRenderer({data, endpoint}) {\n if (data && data.error) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error mb-1'}, 'Search Console error'),\n h('div', {className:'text-[12px] text-on-surface-variant leading-relaxed'}, String((data.error && data.error.message) || data.error))\n )\n );\n }\n\n const fmtNum = (n) => (n == null ? '-' : Math.round(n).toLocaleString());\n const fmtPct = (r) => (r == null ? '-' : (r * 100).toFixed(1) + '%');\n const fmtPos = (p) => (p == null ? '-' : p.toFixed(1));\n\n // --- SITEMAPS ---------------------------------------------------------------\n if (endpoint === 'sitemaps') {\n const maps = Array.isArray(data && data.sitemap) ? data.sitemap : [];\n if (!maps.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No sitemaps submitted.');\n return h('div', {className:'mt-3 space-y-2'},\n maps.map((s, i) => {\n const warn = parseInt(s.warnings || 0, 10);\n const err = parseInt(s.errors || 0, 10);\n const ok = !err && !warn && !s.isPending;\n const accent = err ? '#ffb4ab' : warn ? '#fbbc04' : s.isPending ? '#94a3b8' : '#4edea3';\n return h('a', {\n key:i, href:s.path, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:accent+'18', color:accent}},\n h(Icon, {name:'account_tree', size:22})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, (s.path || '').replace(/^https?:\\\\/\\\\//,'')),\n h('div', {className:'flex flex-wrap items-center gap-2 mt-1.5 text-[10px]'},\n s.type && h('span', {className:'px-1.5 py-0.5 rounded bg-surface-container-high text-on-surface-variant uppercase font-semibold'}, s.type),\n s.lastSubmitted && h('span', {className:'text-text-muted font-mono-data'}, 'submitted ' + fmtRelative(s.lastSubmitted)),\n s.isPending && h('span', {className:'px-1.5 py-0.5 rounded bg-surface-container-high text-text-muted font-semibold'}, 'PENDING'),\n err > 0 && h('span', {className:'px-1.5 py-0.5 rounded text-error font-semibold', style:{background:'#ffb4ab18'}}, err + ' error' + (err===1?'':'s')),\n warn > 0 && h('span', {className:'px-1.5 py-0.5 rounded text-tertiary font-semibold', style:{background:'#fbbc0418'}}, warn + ' warning' + (warn===1?'':'s')),\n ok && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-secondary font-semibold', style:{background:'#4edea318'}}, h(Icon,{name:'check_circle',size:10,filled:true}), 'HEALTHY'),\n Array.isArray(s.contents) && s.contents.length > 0 && h('span', {className:'text-text-muted font-mono-data'}, s.contents.length + ' content type' + (s.contents.length===1?'':'s'))\n )\n ),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n );\n })\n );\n }\n\n // --- SEARCH ANALYTICS / PAGES -----------------------------------------------\n const isPages = endpoint === 'pages';\n const rows = Array.isArray(data && data.rows) ? data.rows : [];\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No search data for this range.');\n\n const sorted = rows.slice().sort((a, b) => (b.clicks || 0) - (a.clicks || 0));\n const maxClicks = Math.max(1, ...sorted.map(r => r.clicks || 0));\n const totals = sorted.reduce((acc, r) => {\n acc.clicks += (r.clicks || 0);\n acc.impressions += (r.impressions || 0);\n return acc;\n }, {clicks:0, impressions:0});\n const avgCtr = totals.impressions ? totals.clicks / totals.impressions : 0;\n const avgPos = sorted.reduce((s, r) => s + (r.position || 0), 0) / sorted.length;\n\n const chip = (label, value, icon, color) => h('div', {className:'glass-panel rounded-xl p-md flex flex-col gap-1'},\n h('div', {className:'flex items-center gap-1.5 text-[10px] uppercase tracking-widest text-text-muted'}, h(Icon,{name:icon,size:12,className:'text-secondary'}), label),\n h('div', {className:'text-[22px] font-bold font-mono-data', style:color?{color}:null}, value)\n );\n\n return h('div', {className:'mt-3 space-y-4'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(140px, 1fr))'}},\n chip('Total Clicks', fmtNum(totals.clicks), 'ads_click'),\n chip('Impressions', fmtNum(totals.impressions), 'visibility'),\n chip('Avg CTR', fmtPct(avgCtr), 'percent'),\n chip('Avg Position', fmtPos(avgPos), 'leaderboard')\n ),\n h('div', {className:'glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'},\n h(Icon,{name: isPages ? 'description' : 'search', size:16, className:'text-secondary'}),\n isPages ? 'Top Pages' : 'Top Queries'\n ),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, sorted.length + (isPages ? ' pages' : ' queries'))\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[520px] overflow-y-auto'},\n sorted.map((r, i) => {\n const key = (r.keys && r.keys[0]) || '';\n const pct = ((r.clicks || 0) / maxClicks) * 100;\n const label = isPages ? key.replace(/^https?:\\\\/\\\\//,'') : key;\n const inner = h('div', {className:'flex items-center gap-3 p-3 hover:bg-white/[0.02] transition-colors' + (isPages ? ' cursor-pointer group' : '')},\n h('div', {className:'w-6 text-center text-[11px] font-mono-data text-text-muted flex-shrink-0'}, i + 1),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-semibold text-on-surface truncate flex items-center gap-1.5'},\n label || '(empty)',\n isPages && h(Icon, {name:'open_in_new', size:12, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n h('div', {className:'h-1.5 mt-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full transition-all', style:{width: Math.max(2, pct) + '%', background:'linear-gradient(90deg,#4edea3,#6ffbbe)'}})\n )\n ),\n h('div', {className:'flex items-center gap-4 flex-shrink-0 text-right'},\n h('div', {className:'w-14'},\n h('div', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtNum(r.clicks)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'clicks')\n ),\n h('div', {className:'w-16 hidden sm:block'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface-variant'}, fmtNum(r.impressions)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'impr')\n ),\n h('div', {className:'w-12'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface'}, fmtPct(r.ctr)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'ctr')\n ),\n h('div', {className:'w-12'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface'}, fmtPos(r.position)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'pos')\n )\n )\n );\n return isPages && key\n ? h('a', {key:i, href:key, target:'_blank', rel:'noreferrer', className:'block'}, inner)\n : h('div', {key:i}, inner);\n })\n )\n )\n );\n}\n`;\n",[],[359,632],"d4ae5f44-5b4e-11f1-a491-ee650bb23af1","gmail.ts","d4ae603e-5b4e-11f1-a491-ee650bb23af1",527,["D",1780052955889],"// GmailRenderer — per-integration renderer for the 'gmail' integration.\nexport const RENDERER = `\nexport function GmailRenderer({data, endpoint}) {\n if (endpoint === 'profile' && data?.emailAddress) {\n return h('div', {className:'glass-panel rounded-xl p-lg flex items-center gap-4 mt-3'},\n h('div', {\n className:'w-14 h-14 rounded-full flex items-center justify-center text-on-primary font-bold text-[18px]',\n style:{background:'linear-gradient(135deg,#ea4335,#c0c1ff)'}\n }, initialsOf(data.emailAddress)),\n h('div', {className:'flex-1'},\n h('h3', {className:'font-bold text-[18px]'}, data.emailAddress),\n h('div', {className:'flex gap-4 mt-1 text-[12px] text-text-muted font-mono-data'},\n h('span', null, h(Icon,{name:'mail',size:12,className:'inline mr-1'}), data.messagesTotal?.toLocaleString() + ' messages'),\n h('span', null, h(Icon,{name:'forum',size:12,className:'inline mr-1'}), data.threadsTotal?.toLocaleString() + ' threads'),\n h('span', null, 'History ID: ' + data.historyId)\n )\n )\n );\n }\n if (endpoint === 'labels' && Array.isArray(data?.labels)) {\n const SYSTEM = data.labels.filter(l => l.type === 'system');\n const USER = data.labels.filter(l => l.type !== 'system');\n const chip = (l) => h('span', {key:l.id, className:'px-2.5 py-1 rounded-full text-[11px] font-medium bg-surface-container-high text-on-surface-variant border border-border-subtle'},\n l.name, ' ', l.messagesUnread > 0 && h('span', {className:'ml-1 px-1.5 rounded bg-primary/20 text-primary text-[10px]'}, l.messagesUnread)\n );\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 space-y-md'},\n h('div', null,\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-2'}, 'System Labels'),\n h('div', {className:'flex flex-wrap gap-1.5'}, SYSTEM.map(chip))\n ),\n USER.length > 0 && h('div', null,\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-2'}, 'Your Labels'),\n h('div', {className:'flex flex-wrap gap-1.5'}, USER.map(chip))\n )\n );\n }\n const messages = Array.isArray(data?._hydrated) ? data._hydrated : [];\n if (!messages.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No messages.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex justify-between items-center bg-surface-container-lowest/50'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'},\n h(Icon,{name:'inbox',size:18,className:'text-error'}),\n (data.resultSizeEstimate || messages.length).toLocaleString() + ' messages'\n ),\n h('span', {className:'text-[10px] font-mono-data text-text-muted'}, 'showing ' + messages.length)\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n messages.map(m => {\n const fromName = (m.from || '').replace(/<.*>/,'').trim() || m.from;\n const c = colorFromName(fromName);\n return h('div', {\n key: m.id,\n className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors cursor-pointer ' + (m.unread ? 'border-l-2 border-l-primary' : '')\n },\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(fromName)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className: 'text-[13px] truncate ' + (m.unread?'font-bold text-on-surface':'text-on-surface-variant')}, fromName),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.internalDate ? new Date(parseInt(m.internalDate,10)).toISOString() : m.date))\n ),\n h('div', {className:'text-[13px] truncate ' + (m.unread?'font-semibold text-on-surface':'text-on-surface-variant')}, m.subject || '(no subject)'),\n m.snippet && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, m.snippet),\n (m.starred || (m.labelIds||[]).filter(l => !['INBOX','UNREAD','CATEGORY_PERSONAL','SENT','STARRED'].includes(l)).length > 0) && h('div', {className:'flex flex-wrap gap-1 mt-1.5'},\n m.starred && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-tertiary bg-tertiary/10 border border-tertiary/30'},\n h(Icon,{name:'star',size:10,filled:true}), 'STARRED'\n ),\n (m.labelIds||[]).filter(l => !['INBOX','UNREAD','CATEGORY_PERSONAL','SENT','STARRED'].includes(l)).slice(0,3).map(l =>\n h('span', {key:l, className:'px-1.5 py-0.5 rounded text-[9px] font-semibold text-on-surface-variant bg-surface-container-high'}, l.replace('CATEGORY_','').replace(/_/g,' '))\n )\n )\n )\n );\n })\n )\n );\n}\n\n// --- DRIVE -------------------------------------------------------------------\nconst MIME_ICON = {\n 'application/vnd.google-apps.folder': {icon:'folder', color:'#fbbc04'},\n 'application/vnd.google-apps.document': {icon:'description', color:'#4285f4'},\n 'application/vnd.google-apps.spreadsheet': {icon:'table_chart', color:'#0f9d58'},\n 'application/vnd.google-apps.presentation': {icon:'co_present', color:'#f4b400'},\n 'application/vnd.google-apps.form': {icon:'list_alt', color:'#673ab7'},\n 'application/pdf': {icon:'picture_as_pdf', color:'#ea4335'},\n 'application/zip': {icon:'folder_zip', color:'#5f6368'},\n 'image/png': {icon:'image', color:'#34a853'},\n 'image/jpeg': {icon:'image', color:'#34a853'},\n 'video/mp4': {icon:'movie', color:'#ea4335'},\n};\nconst driveIconFor = (mime) => MIME_ICON[mime] || (mime?.startsWith('image/') ? MIME_ICON['image/png'] : mime?.startsWith('video/') ? MIME_ICON['video/mp4'] : {icon:'insert_drive_file', color:'#94a3b8'});\n`;\n",[],[359,640],"d43df1aa-5b4e-11f1-90b4-ee650bb23af1","ga4.ts","d43df27c-5b4e-11f1-90b4-ee650bb23af1",526,["D",1780052955152],"// GA4Renderer — per-integration renderer for the 'ga4' integration.\nexport const RENDERER = `\nexport function GA4Renderer({data, endpoint}) {\n if (data && data.error) {\n const msg = (data.error && data.error.message) || data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error mb-1'}, 'Google Analytics error'),\n h('div', {className:'text-[12px] text-on-surface-variant leading-relaxed'}, String(msg))\n )\n );\n }\n const fmtN = (v) => { const n = parseFloat(v) || 0; return n >= 1e6 ? (n/1e6).toFixed(1)+'M' : n >= 1e3 ? (n/1e3).toFixed(1)+'K' : String(Math.round(n*100)/100); };\n const isRate = (k) => /rate|percent/i.test(k || '');\n const fmtMetric = (k, v) => isRate(k) ? (parseFloat(v||'0')*100).toFixed(1)+'%' : fmtN(v);\n const parseRows = (d) => {\n if (!d || !d.rows || !d.rows.length) return [];\n const dh = (d.dimensionHeaders || []).map(x => x.name);\n const mh = (d.metricHeaders || []).map(x => x.name);\n return d.rows.map(r => {\n const o = {};\n dh.forEach((k,i) => { o[k] = (r.dimensionValues || [])[i] && r.dimensionValues[i].value; });\n mh.forEach((k,i) => { o[k] = (r.metricValues || [])[i] && r.metricValues[i].value; });\n return o;\n });\n };\n const empty = (t) => h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, t);\n const dimNames = (data && data.dimensionHeaders || []).map(x => x.name);\n const metNames = (data && data.metricHeaders || []).map(x => x.name);\n\n // --- KPI (two date ranges; current then previous metrics) -------------------\n if (endpoint === 'kpi') {\n const row = data && data.rows && data.rows[0];\n if (!row || !metNames.length) return empty('No KPI data for this range.');\n const vals = row.metricValues || [];\n const n = metNames.length;\n const META = {sessions:'#c0c1ff', activeUsers:'#4edea3', newUsers:'#d0bcff', screenPageViews:'#8083ff', engagementRate:'#6ffbbe', bounceRate:'#ffb4ab'};\n const labelOf = (k) => k.replace(/([A-Z])/g,' $1').replace(/^./, c => c.toUpperCase());\n return h('div', {className:'grid gap-3 mt-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(155px, 1fr))'}},\n metNames.map((k, idx) => {\n const color = META[k] || '#c0c1ff';\n const cur = parseFloat((vals[idx] && vals[idx].value) || '0');\n const prev = (idx + n) < vals.length ? parseFloat((vals[idx+n] && vals[idx+n].value) || '0') : null;\n const delta = (prev != null && prev !== 0) ? ((cur - prev) / prev) * 100 : null;\n const up = delta != null && delta >= 0;\n const good = isRate(k) ? (/bounce/i.test(k) ? !up : up) : up;\n return h('div', {key:k, className:'glass-panel rounded-xl p-4', style:{borderColor:color+'28'}},\n h('div', {className:'text-[10px] text-text-muted uppercase tracking-wider mb-2'}, labelOf(k)),\n h('div', {className:'flex items-baseline gap-2 flex-wrap'},\n h('div', {className:'text-[26px] font-extrabold text-on-surface leading-none font-mono-data'}, fmtMetric(k, String(cur))),\n delta != null && h('span', {className:'inline-flex items-center gap-0.5 text-[11px] font-bold', style:{color: good ? '#4edea3' : '#ffb4ab'}},\n h(Icon, {name: up ? 'trending_up' : 'trending_down', size:12}), Math.abs(delta).toFixed(0)+'%')\n ),\n h('div', {className:'mt-2 h-0.5 rounded', style:{background:color, opacity:0.35}})\n );\n })\n );\n }\n\n // --- TRAFFIC OVER TIME (date dimension trend table) -------------------------\n if (endpoint === 'traffic-over-time') {\n const rows = parseRows(data);\n if (!rows.length) return empty('No traffic data for this range.');\n const max = Math.max(1, ...rows.map(r => parseFloat(r.sessions || 0)));\n const total = rows.reduce((s,r) => s + (parseFloat(r.sessions)||0), 0);\n return h('div', {className:'mt-3 glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:'show_chart',size:16,className:'text-secondary'}), 'Sessions Over Time'),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, fmtN(total)+' total')\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[440px] overflow-y-auto'},\n rows.map((r,i) => {\n const d = String(r.date || '');\n const lbl = d.length === 8 ? d.slice(4,6)+'/'+d.slice(6,8) : d;\n const pct = (parseFloat(r.sessions||0) / max) * 100;\n return h('div', {key:i, className:'flex items-center gap-3 px-3 py-1.5'},\n h('div', {className:'w-12 text-[11px] font-mono-data text-text-muted flex-shrink-0'}, lbl),\n h('div', {className:'flex-1 h-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#c0c1ff,#8083ff)'}})),\n h('div', {className:'w-14 text-right text-[12px] font-bold font-mono-data text-on-surface'}, fmtN(r.sessions)),\n h('div', {className:'w-12 text-right text-[11px] font-mono-data text-text-muted'}, fmtN(r.activeUsers))\n );\n })\n )\n );\n }\n\n // --- REALTIME (active users by country, last 30 min) ------------------------\n if (endpoint === 'realtime') {\n const rows = parseRows(data);\n const total = rows.reduce((s,r) => s + (parseFloat(r.activeUsers)||0), 0);\n return h('div', {className:'mt-3 space-y-3'},\n h('div', {className:'glass-panel rounded-xl p-4 flex items-center gap-3', style:{borderColor:'#4edea328'}},\n h('span', {className:'w-2.5 h-2.5 rounded-full animate-pulse', style:{background:'#4edea3'}}),\n h('div', {className:'text-[30px] font-extrabold font-mono-data text-on-surface leading-none'}, fmtN(total)),\n h('div', {className:'text-[11px] text-text-muted uppercase tracking-wider'}, 'active users now')\n ),\n rows.length ? h('div', {className:'glass-panel rounded-xl divide-y divide-border-subtle/50'},\n rows.map((r,i) => h('div', {key:i, className:'flex items-center justify-between px-3 py-2'},\n h('span', {className:'text-[13px] text-on-surface'}, r[dimNames[0]] || '(unknown)'),\n h('span', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtN(r.activeUsers))\n ))\n ) : empty('No active users in the last 30 minutes.')\n );\n }\n\n // --- FUNNEL (alpha runFunnelReport) -----------------------------------------\n if (endpoint === 'funnel') {\n const ft = data && (data.funnelTable || data.subFunnelTables && data.subFunnelTables[0]);\n const frows = ft && ft.rows || [];\n if (!frows.length) return empty('No funnel data. Verify event names / property ID.');\n const cnt = (r) => parseFloat((r.metricValues && r.metricValues[0] && r.metricValues[0].value) || 0);\n const dv = (r) => ((r.dimensionValues || []).map(x => x.value).join(' / ')) || 'Step';\n const top = Math.max(1, ...frows.map(cnt));\n return h('div', {className:'mt-3 space-y-2'},\n frows.map((r,i) => {\n const c = cnt(r), pct = (c/top)*100;\n return h('div', {key:i, className:'glass-panel rounded-xl p-3'},\n h('div', {className:'flex justify-between items-center mb-1.5'},\n h('span', {className:'text-[13px] font-semibold text-on-surface'}, (i+1)+'. '+dv(r)),\n h('span', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtN(c)+' ('+pct.toFixed(0)+'%)')),\n h('div', {className:'h-2 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#4edea3,#6ffbbe)'}}))\n );\n })\n );\n }\n\n // --- RANKED / TABULAR: user-acquisition, report, page-views, top-events, device-breakdown\n const rows = parseRows(data);\n if (!rows.length) return empty('No data for this range.');\n const isPages = endpoint === 'page-views';\n const sortKey = metNames.find(k => /sessions|eventCount|screenPageViews/i.test(k)) || metNames[0];\n const sorted = rows.slice().sort((a,b) => (parseFloat(b[sortKey])||0) - (parseFloat(a[sortKey])||0));\n const max = Math.max(1, ...sorted.map(r => parseFloat(r[sortKey])||0));\n const titleMap = {'user-acquisition':'Acquisition by Channel','report':'Traffic Report','page-views':'Top Pages','top-events':'Top Events','device-breakdown':'Devices'};\n const iconMap = {'user-acquisition':'campaign','report':'table_chart','page-views':'description','top-events':'bolt','device-breakdown':'devices'};\n const totals = {};\n metNames.forEach(k => { totals[k] = sorted.reduce((s,r) => s + (parseFloat(r[k])||0), 0); });\n return h('div', {className:'mt-3 space-y-4'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(140px, 1fr))'}},\n metNames.slice(0,4).map(k => h('div', {key:k, className:'glass-panel rounded-xl p-md flex flex-col gap-1'},\n h('div', {className:'text-[10px] uppercase tracking-widest text-text-muted'}, k.replace(/([A-Z])/g,' $1')),\n h('div', {className:'text-[22px] font-bold font-mono-data text-on-surface'}, isRate(k) ? (totals[k]/sorted.length*100).toFixed(1)+'%' : fmtN(totals[k]))\n ))\n ),\n h('div', {className:'glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:iconMap[endpoint]||'analytics',size:16,className:'text-secondary'}), titleMap[endpoint] || 'Report'),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, sorted.length + ' rows')\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[520px] overflow-y-auto'},\n sorted.map((r,i) => {\n const key = dimNames.map(d => r[d]).filter(Boolean).join(' / ') || '(empty)';\n const label = isPages ? key.replace(/^https?:\\\\/\\\\//,'') : key;\n const pct = (parseFloat(r[sortKey])||0) / max * 100;\n const inner = h('div', {className:'flex items-center gap-3 p-3 hover:bg-white/[0.02] transition-colors' + (isPages ? ' group cursor-pointer' : '')},\n h('div', {className:'w-6 text-center text-[11px] font-mono-data text-text-muted flex-shrink-0'}, i+1),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-semibold text-on-surface truncate flex items-center gap-1.5'}, label,\n isPages && h(Icon,{name:'open_in_new',size:12,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})),\n h('div', {className:'h-1.5 mt-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#c0c1ff,#8083ff)'}}))\n ),\n h('div', {className:'flex items-center gap-4 flex-shrink-0 text-right'},\n metNames.slice(0,3).map(k => h('div', {key:k, className:'w-14'},\n h('div', {className:'text-[12px] font-bold font-mono-data text-on-surface'}, fmtMetric(k, r[k])),\n h('div', {className:'text-[9px] text-text-muted uppercase truncate'}, k.replace(/([A-Z]).*/,'').slice(0,6) || k.slice(0,6))\n ))\n )\n );\n return isPages && r[dimNames[0]] ? h('a', {key:i, href:r[dimNames[0]], target:'_blank', rel:'noreferrer', className:'block'}, inner) : h('div', {key:i}, inner);\n })\n )\n )\n );\n}\n`;\n",[],[359,648],"d392f818-5b4e-11f1-b8d2-ee650bb23af1","excel.ts","d392f8c2-5b4e-11f1-b8d2-ee650bb23af1",525,["D",1780052954028],"// ExcelRenderer — per-integration renderer for the 'excel' integration.\nexport const RENDERER = `\nexport function ExcelRenderer({data, endpoint}) {\n const GREEN = '#217346';\n if (data && data.error) {\n const e = data.error;\n const msg = (e && typeof e === 'object') ? (e.message || e.code || JSON.stringify(e)) : String(e);\n const code = (e && typeof e === 'object' && e.code) ? e.code : null;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h('div', {className:'w-9 h-9 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#21734618',color:GREEN}}, h(Icon,{name:'table_chart',size:20})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-error flex items-center gap-1.5'}, h(Icon,{name:'error',size:14}), 'Excel error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1 break-words'}, msg),\n code && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-1'}, code)\n )\n );\n }\n if (endpoint === 'range') {\n const rows = (data && Array.isArray(data.values)) ? data.values : [];\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No data in used range.');\n const header = rows[0] || [];\n const bodyRows = rows.slice(1);\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[12px] font-bold flex items-center gap-2'},\n h('div', {className:'w-5 h-5 rounded flex items-center justify-center flex-shrink-0', style:{background:GREEN+'22',color:GREEN}}, h(Icon,{name:'grid_on',size:13})),\n data && data.address ? data.address : 'Used range'\n ),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, rows.length + ' rows - ' + header.length + ' cols')\n ),\n h('div', {className:'overflow-x-auto max-h-[480px]'},\n h('table', {className:'w-full text-[12px] border-collapse'},\n h('thead', null,\n h('tr', {className:'sticky top-0'},\n h('th', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-b border-r border-border-subtle bg-surface-container-lowest', style:{width:40}}, ''),\n header.map((c, j) => h('th', {key:j, className:'px-3 py-1.5 text-left font-bold border-b border-border-subtle text-on-surface', style:{background:GREEN+'14'}}, String(c == null ? '' : c)))\n )\n ),\n h('tbody', null,\n bodyRows.slice(0, 200).map((row, i) => h('tr', {key:i, className:'border-b border-border-subtle/50 hover:bg-white/[0.02]'},\n h('td', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-r border-border-subtle/50', style:{width:40}}, i+2),\n (Array.isArray(row) ? row : []).map((cell, j) => h('td', {key:j, className:'px-3 py-1.5 whitespace-nowrap'}, String(cell == null ? '' : cell)))\n ))\n )\n )\n )\n );\n }\n if (endpoint === 'worksheets') {\n const sheets = (data && Array.isArray(data.value)) ? data.value : [];\n if (!sheets.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No worksheets.');\n const ordered = sheets.slice().sort((a, b) => (a.position || 0) - (b.position || 0));\n return h('div', {className:'glass-panel rounded-xl p-md mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3 flex items-center gap-1.5'},\n h(Icon,{name:'tab',size:12}), sheets.length + ' Worksheet' + (sheets.length === 1 ? '' : 's')\n ),\n h('div', {className:'flex flex-wrap gap-2'},\n ordered.map((s, i) => {\n const hidden = s.visibility && s.visibility !== 'Visible';\n return h('div', {\n key: s.id || i,\n className:'flex items-center gap-2 px-3 py-2 rounded-lg border bg-surface-container-lowest ' + (hidden ? 'border-border-subtle/60 opacity-60' : 'border-border-subtle'),\n style: hidden ? {} : {borderColor: GREEN + '55'}\n },\n h('div', {className:'w-2 h-2 rounded-sm flex-shrink-0', style:{background: hidden ? '#888' : GREEN}}),\n h('div', {className:'flex flex-col min-w-0'},\n h('span', {className:'text-[13px] font-bold text-on-surface truncate'}, s.name || '(unnamed)'),\n h('span', {className:'text-[9px] text-text-muted font-mono-data'},\n 'pos ' + (s.position != null ? s.position : '-') + (hidden ? ' - ' + s.visibility : '')\n )\n )\n );\n })\n )\n );\n }\n // Default: workbooks\n const files = (data && Array.isArray(data.value)) ? data.value : [];\n if (!files.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No Excel workbooks found.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n files.map(f => h('a', {\n key: f.id, href: f.webUrl, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#21734618',color:'#217346'}},\n h(Icon,{name:'table_chart',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, f.name || '(unnamed)'),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted flex-wrap'},\n f.id && h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), String(f.id).slice(0, 8) + '-'\n ),\n h('span', {className:'font-mono-data'}, fmtRelative(f.lastModifiedDateTime)),\n f.size != null && h('span', {className:'font-mono-data text-on-surface-variant'}, fmtFileSize(f.size))\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,656],"d2e5ab22-5b4e-11f1-8887-ee650bb23af1","drive.ts","d2e5abd6-5b4e-11f1-8887-ee650bb23af1",524,["D",1780052952897],"// DriveRenderer — per-integration renderer for the 'drive' integration.\nexport const RENDERER = `\nexport function DriveRenderer({data, endpoint}) {\n if (endpoint === 'about' && data?.user) {\n const q = data.storageQuota || {};\n const usedGB = (parseInt(q.usage||0,10) / 1073741824).toFixed(2);\n const limitGB = q.limit ? (parseInt(q.limit,10) / 1073741824).toFixed(0) : '-';\n const pct = q.limit ? Math.min(100, (parseInt(q.usage||0,10) / parseInt(q.limit,10)) * 100) : 0;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-4 mb-4'},\n data.user.photoLink ? h('img',{src:data.user.photoLink, className:'w-14 h-14 rounded-full border border-border-subtle'}) : null,\n h('div', null,\n h('h3', {className:'font-bold text-[18px]'}, data.user.displayName),\n h('div', {className:'text-[12px] text-text-muted'}, data.user.emailAddress)\n )\n ),\n h('div', {className:'space-y-2'},\n h('div', {className:'flex justify-between text-[12px]'},\n h('span', {className:'text-on-surface-variant'}, 'Storage'),\n h('span', {className:'font-mono-data font-bold'}, usedGB + ' GB / ' + limitGB + ' GB')\n ),\n h('div', {className:'h-2 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full transition-all', style:{width: pct + '%', background: pct > 90 ? '#ffb4ab' : '#4edea3'}})\n )\n )\n );\n }\n const files = data?.files || [];\n if (!files.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No files.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(220px, 1fr))'}},\n files.map(f => {\n const ico = driveIconFor(f.mimeType);\n const owner = f.owners?.[0];\n return h('a', {\n key: f.id, href: f.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex flex-col gap-2 hover:border-primary/30 transition-all group'\n },\n h('div', {className:'flex items-start justify-between gap-2'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background: ico.color+'18', color: ico.color}},\n h(Icon,{name: ico.icon, size:22})\n ),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, f.name),\n h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, f.mimeType?.replace('application/vnd.google-apps.','').replace('application/','') || '-')\n ),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted pt-2 border-t border-border-subtle'},\n h('span', {className:'truncate flex items-center gap-1'},\n owner?.photoLink ? h('img',{src:owner.photoLink, className:'w-4 h-4 rounded-full'}) : null,\n owner?.displayName || '-'\n ),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(f.modifiedTime)),\n f.size && h('span', {className:'font-mono-data text-on-surface-variant flex-shrink-0'}, fmtFileSize(f.size))\n ),\n (f.shared || f.starred) && h('div', {className:'flex gap-1 flex-wrap'},\n f.starred && h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-tertiary/10 text-tertiary'}, h(Icon,{name:'star',size:9,filled:true}), 'STARRED'),\n f.shared && h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-secondary/10 text-secondary'}, h(Icon,{name:'group',size:9}), 'SHARED')\n )\n );\n })\n )\n );\n}\n`;\n",[],[359,664],"24f6f7ae-5b54-11f1-a375-ee650bb23af1","docs.ts","d2841768-5b4e-11f1-a491-ee650bb23af1",549,["D",1780055238021],"// DocsRenderer — per-integration renderer for the 'docs' integration.\nexport const RENDERER = `\nexport function DocsRenderer({data, endpoint}) {\n if (endpoint === 'document' && data?.body) {\n const blocks = data.body.content || [];\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-3 mb-4 pb-3 border-b border-border-subtle'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#4285f418',color:'#4285f4'}},\n h(Icon,{name:'description',size:22})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('h3', {className:'text-[16px] font-bold truncate'}, data.title),\n h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, blocks.length + ' content blocks - revision ' + (data.revisionId?.slice(0,8) || '-'))\n )\n ),\n h('div', {className:'space-y-2 max-h-[480px] overflow-y-auto text-[13px] text-on-surface-variant leading-relaxed'},\n blocks.slice(0,30).map((b,i) => {\n const text = (b.paragraph?.elements || []).map(e => e.textRun?.content || '').join('');\n if (!text.trim()) return null;\n const isH1 = b.paragraph?.paragraphStyle?.namedStyleType === 'HEADING_1';\n const isH2 = b.paragraph?.paragraphStyle?.namedStyleType === 'HEADING_2';\n return h('p', {key:i, className: isH1?'text-[18px] font-bold text-on-surface':isH2?'text-[15px] font-bold text-on-surface mt-2':'text-[13px]'}, text);\n })\n )\n );\n }\n const docs = data?.files || [];\n if (!docs.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No documents.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(260px, 1fr))'}},\n docs.map(d => h('a', {\n key:d.id, href:d.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#4285f418',color:'#4285f4'}},\n h(Icon,{name:'description',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, d.name),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted'},\n h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), (d.id||'').slice(0,8) + '-'\n ),\n h('span', null, '-'),\n h('span', null, fmtRelative(d.modifiedTime))\n ),\n d.owners?.[0] && h('div', {className:'text-[10px] text-on-surface-variant truncate mt-1 flex items-center gap-1'},\n h(Icon,{name:'person',size:11}), d.owners[0].displayName\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,672],"d224620a-5b4e-11f1-95e0-ee650bb23af1","calendar.ts","d22462dc-5b4e-11f1-95e0-ee650bb23af1",522,["D",1780052951618],"// CalendarRenderer — per-integration renderer for the 'calendar' integration.\nexport const RENDERER = `\nexport function CalendarRenderer({data, endpoint}) {\n if (endpoint === 'calendars' && Array.isArray(data?.items)) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 space-y-2'},\n h('h3', {className:'font-bold text-[15px] mb-2'}, data.items.length + ' Calendars'),\n data.items.map(c => h('div', {key:c.id, className:'flex items-center gap-3 p-2.5 rounded-lg bg-surface-container/40'},\n h('div', {className:'w-3 h-3 rounded flex-shrink-0', style:{background: c.backgroundColor || '#c0c1ff'}}),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold truncate'}, c.summary, c.primary && h('span', {className:'ml-2 text-[9px] text-primary uppercase'}, 'PRIMARY')),\n h('div', {className:'text-[11px] text-text-muted truncate'}, c.id)\n ),\n h('span', {className:'text-[10px] text-on-surface-variant uppercase'}, c.accessRole)\n ))\n );\n }\n const events = data?.items || [];\n if (!events.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No events scheduled.');\n const groups = {};\n events.forEach(e => {\n const startISO = e.start?.dateTime || e.start?.date;\n const dKey = startISO ? new Date(startISO).toDateString() : 'undated';\n if (!groups[dKey]) groups[dKey] = [];\n groups[dKey].push(e);\n });\n const todayStr = new Date().toDateString();\n const tomStr = new Date(Date.now()+86400000).toDateString();\n const dateLabel = (d) => d === todayStr ? 'TODAY' : d === tomStr ? 'TOMORROW' : new Date(d).toLocaleDateString('en-US',{weekday:'long',month:'short',day:'numeric'}).toUpperCase();\n return h('div', {className:'mt-3 space-y-4'},\n Object.entries(groups).map(([dKey, evs]) => h('div', {key:dKey},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-secondary mb-2 sticky top-0 bg-background/80 backdrop-blur-sm py-1'},\n dateLabel(dKey), h('span', {className:'ml-2 text-text-muted'}, '- ' + evs.length + ' event' + (evs.length===1?'':'s'))\n ),\n h('div', {className:'space-y-2'},\n evs.map(e => {\n const startT = e.start?.dateTime ? new Date(e.start.dateTime).toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : 'All-day';\n const endT = e.end?.dateTime ? new Date(e.end.dateTime).toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : '';\n const attendees = (e.attendees||[]).filter(a => !a.self);\n return h('a', {\n key:e.id, href:e.htmlLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-lg p-3 flex items-start gap-3 hover:border-secondary/30 transition-all group'\n },\n h('div', {className:'flex flex-col items-center justify-center w-16 flex-shrink-0 pr-2 border-r border-border-subtle'},\n h('div', {className:'text-[12px] font-bold text-on-surface font-mono-data'}, startT),\n endT && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, '- ' + endT)\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, e.summary || '(no title)'),\n e.location && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1'},\n h(Icon,{name:'place',size:11}), e.location\n ),\n e.description && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, e.description.replace(/<[^>]+>/g,'').slice(0,80)),\n attendees.length > 0 && h('div', {className:'flex items-center gap-1 mt-1.5 text-[10px] text-text-muted'},\n h(Icon,{name:'group',size:11}),\n attendees.length + ' attendee' + (attendees.length===1?'':'s'),\n e.hangoutLink && h('span', {className:'ml-1 text-secondary'}, '- Meet')\n )\n ),\n e.status === 'confirmed' && h('div', {className:'w-1.5 h-1.5 rounded-full bg-secondary mt-2 flex-shrink-0'})\n );\n })\n )\n ))\n );\n}\n`;\n",[],[359,680],"d19debbc-5b4e-11f1-a8fd-ee650bb23af1","ahrefs.ts","d19ded24-5b4e-11f1-a8fd-ee650bb23af1",521,["D",1780052950747],"// AhrefsRenderer — per-integration renderer for the 'ahrefs' integration.\nexport const RENDERER = `\nexport function AhrefsRenderer({data, endpoint}) {\n const AH = '#ff6924';\n if (!data) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No data.');\n if (data.error) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon,{name:'error',size:18,className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error'}, 'Ahrefs API error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1'}, String(data.error))\n )\n );\n\n const num = (n) => (n == null || n === '') ? '-' : Number(n).toLocaleString();\n const compact = (n) => {\n const v = Number(n) || 0;\n if (v >= 1e6) return (v/1e6).toFixed(1).replace(/\\\\.0$/,'') + 'M';\n if (v >= 1e3) return (v/1e3).toFixed(1).replace(/\\\\.0$/,'') + 'K';\n return String(v);\n };\n const drColor = (dr) => dr >= 70 ? '#4edea3' : dr >= 40 ? '#fbbc04' : dr >= 20 ? '#ffa726' : '#94a3b8';\n const hostOf = (u) => { try { return new URL(u).hostname.replace(/^www\\\\./,''); } catch { return u || ''; } };\n\n const chip = (label, value, icon) => h('div', {key:label, className:'glass-panel rounded-xl p-md flex flex-col gap-1 border border-border-subtle'},\n h('div', {className:'flex items-center gap-1.5 text-[10px] uppercase tracking-widest text-text-muted'},\n icon && h(Icon,{name:icon,size:12,style:{color:AH}}), label\n ),\n h('div', {className:'text-[22px] font-bold text-on-surface font-mono-data'}, value)\n );\n const chipGrid = (chips) => h('div', {className:'grid gap-3 mt-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(150px, 1fr))'}}, chips.filter(Boolean));\n\n // -- domain-rating: gauge + rank ------------------------------------------\n if (endpoint === 'domain-rating') {\n const dr = data.domain_rating || data;\n const score = dr.domain_rating != null ? Number(dr.domain_rating) : null;\n const c = drColor(score || 0);\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-6'},\n h('div', {className:'relative w-24 h-24 rounded-full flex items-center justify-center flex-shrink-0',\n style:{background:'conic-gradient('+c+' '+(score||0)*3.6+'deg, rgba(255,255,255,0.07) 0deg)'}},\n h('div', {className:'w-[78px] h-[78px] rounded-full bg-surface-container flex flex-col items-center justify-center'},\n h('span', {className:'text-[28px] font-bold font-mono-data', style:{color:c}}, score != null ? score : '-'),\n h('span', {className:'text-[9px] uppercase tracking-widest text-text-muted'}, 'DR')\n )\n ),\n h('div', {className:'flex-1'},\n h('h3', {className:'text-[16px] font-bold'}, 'Domain Rating'),\n h('div', {className:'text-[12px] text-text-muted mt-0.5'}, 'Backlink strength on a 0-100 scale'),\n dr.ahrefs_rank != null && h('div', {className:'mt-3 inline-flex items-center gap-1.5 px-3 py-1.5 rounded-lg bg-surface-container-high text-[12px] font-mono-data'},\n h(Icon,{name:'trending_up',size:13,style:{color:AH}}), 'Ahrefs Rank #' + num(dr.ahrefs_rank)\n )\n )\n );\n }\n\n // -- site-explorer / backlinks-stats: metric chips ------------------------\n if (endpoint === 'site-explorer' || endpoint === 'backlinks-stats') {\n const m = data.metrics || data;\n const chips = endpoint === 'site-explorer' ? [\n m.org_traffic != null && chip('Org Traffic', compact(m.org_traffic), 'visibility'),\n m.org_keywords != null && chip('Org Keywords', num(m.org_keywords), 'key'),\n m.org_cost != null && chip('Traffic Value', '$' + compact(m.org_cost), 'paid'),\n m.paid_traffic != null && chip('Paid Traffic', compact(m.paid_traffic), 'ads_click'),\n m.paid_keywords != null && chip('Paid Keywords', num(m.paid_keywords), 'campaign'),\n m.paid_cost != null && chip('Paid Cost', '$' + compact(m.paid_cost), 'attach_money')\n ] : [\n m.live != null && chip('Live Backlinks', compact(m.live), 'link'),\n m.all_time != null && chip('All-Time', compact(m.all_time), 'history'),\n m.live_refdomains != null && chip('Live RefDomains', compact(m.live_refdomains), 'public'),\n m.all_time_refdomains != null && chip('All-Time RefDomains', compact(m.all_time_refdomains), 'language')\n ];\n if (!chips.filter(Boolean).length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No metrics returned.');\n return chipGrid(chips);\n }\n\n // -- ranked list builder --------------------------------------------------\n const rankedRows = (rows, opts) => {\n const max = Math.max(1, ...rows.map(opts.metric));\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:opts.icon,size:16,style:{color:AH}}), opts.title),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, rows.length + ' ' + opts.unit)\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n rows.map((r, i) => {\n const pct = Math.max(2, (opts.metric(r) / max) * 100);\n const lbl = opts.label(r);\n return h(opts.href ? 'a' : 'div', Object.assign({key:i,\n className:'block px-3.5 py-2.5 hover:bg-white/[0.02] transition-colors ' + (opts.href ? 'group' : '')},\n opts.href ? {href:opts.href(r), target:'_blank', rel:'noreferrer'} : {}),\n h('div', {className:'flex items-center gap-3'},\n h('span', {className:'text-[10px] text-text-muted font-mono-data w-5 text-right flex-shrink-0'}, i+1),\n opts.badge && h('span', {className:'text-[10px] font-bold font-mono-data px-1.5 py-0.5 rounded flex-shrink-0', style:opts.badge(r)}, opts.badgeText(r)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] truncate text-on-surface ' + (opts.href?'group-hover:text-primary transition-colors':'')}, lbl),\n opts.sub && h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, opts.sub(r))\n ),\n h('div', {className:'text-right flex-shrink-0'},\n h('div', {className:'text-[12px] font-bold font-mono-data text-on-surface'}, opts.metricText(r)),\n opts.metricLabel && h('div', {className:'text-[9px] text-text-muted uppercase'}, opts.metricLabel)\n ),\n opts.href && h(Icon,{name:'open_in_new',size:12,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n h('div', {className:'h-1 mt-1.5 rounded-full bg-surface-container-high overflow-hidden ml-8'},\n h('div', {className:'h-full rounded-full', style:{width:pct+'%', background:opts.barColor ? opts.barColor(r) : AH}})\n )\n );\n })\n )\n );\n };\n\n // -- keywords / organic-keywords ------------------------------------------\n if (endpoint === 'keywords' || endpoint === 'organic-keywords') {\n const rows = data.keywords || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No keywords found.');\n const metric = (r) => Number(r.sum_traffic ?? r.traffic_potential ?? r.volume ?? r.global_volume ?? 0);\n return rankedRows(rows, {\n title:'Keywords', icon:'key', unit:'keywords',\n metric, metricText:(r)=>compact(metric(r)), metricLabel: (rows[0].sum_traffic!=null||rows[0].traffic_potential!=null)?'traffic':'volume',\n label:(r)=>r.keyword || '-',\n sub:(r)=>{\n const parts = [];\n if (r.volume != null) parts.push('Vol ' + compact(r.volume));\n const diff = r.keyword_difficulty ?? r.difficulty;\n if (diff != null) parts.push('KD ' + diff);\n if (r.cpc != null) parts.push('$' + Number(r.cpc).toFixed(2) + ' CPC');\n if (r.best_position != null) parts.push('Pos #' + r.best_position);\n return parts.join(' - ');\n },\n badge:(r)=>{ const d = Number(r.keyword_difficulty ?? r.difficulty ?? 0); const c = drColor(100-d); return {background:c+'22',color:c}; },\n badgeText:(r)=>'KD ' + (r.keyword_difficulty ?? r.difficulty ?? '-')\n });\n }\n\n // -- top-pages ------------------------------------------------------------\n if (endpoint === 'top-pages') {\n const rows = data.pages || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No pages found.');\n const metric = (r) => Number(r.sum_traffic ?? r.value ?? 0);\n return rankedRows(rows, {\n title:'Top Pages', icon:'article', unit:'pages',\n metric, metricText:(r)=>compact(metric(r)), metricLabel:'traffic',\n href:(r)=>r.url, label:(r)=>hostOf(r.url) + (function(){ try { return new URL(r.url).pathname; } catch { return ''; } })(),\n sub:(r)=>{\n const parts = [];\n if (r.top_keyword) parts.push('Top: ' + r.top_keyword + (r.top_keyword_best_position!=null?' (#'+r.top_keyword_best_position+')':''));\n if (r.keywords != null) parts.push(compact(r.keywords) + ' keywords');\n return parts.join(' - ');\n }\n });\n }\n\n // -- backlinks / referring-domains ----------------------------------------\n if (endpoint === 'backlinks' || endpoint === 'referring-domains') {\n const rows = data.backlinks || data.refdomains || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No backlinks found.');\n const isRef = endpoint === 'referring-domains';\n const dr = (r) => Number(r.domain_rating ?? r.domain_rating_source ?? 0);\n return rankedRows(rows, {\n title: isRef ? 'Referring Domains' : 'Backlinks', icon: isRef ? 'public' : 'link', unit: isRef ? 'domains' : 'backlinks',\n metric: dr, metricText:(r)=>dr(r) || '-', metricLabel:'DR', barColor:(r)=>drColor(dr(r)),\n href:(r)=> isRef ? ('https://' + (r.domain||'')) : r.url_from,\n label:(r)=> isRef ? (r.domain || '-') : (r.title || r.name_source || hostOf(r.url_from)),\n sub:(r)=>{\n if (isRef) {\n const parts = [];\n if (r.dofollow_links != null) parts.push(compact(r.dofollow_links) + ' dofollow links');\n if (r.traffic_domain != null) parts.push(compact(r.traffic_domain) + ' traffic');\n return parts.join(' - ');\n }\n const parts = [];\n if (r.anchor) parts.push('\"' + r.anchor.slice(0,50) + '\"');\n parts.push(hostOf(r.url_from));\n return parts.join(' - ');\n },\n badge:(r)=> (!isRef && r.is_dofollow != null) ? {background:(r.is_dofollow?'#4edea3':'#94a3b8')+'22', color:(r.is_dofollow?'#4edea3':'#94a3b8')} : null,\n badgeText:(r)=> r.is_dofollow ? 'FOLLOW' : 'NOFOLLOW'\n });\n }\n\n // -- fallback -------------------------------------------------------------\n return h('div', {className:'glass-panel rounded-xl p-md mt-3 text-[12px] text-text-muted font-mono-data overflow-x-auto'},\n h('pre', null, JSON.stringify(data, null, 2).slice(0, 1200))\n );\n}\n`;\n",[],[359,688],"d0f3f4b8-5b4e-11f1-8d86-ee650bb23af1","_preamble.ts","d0f3f576-5b4e-11f1-8d86-ee650bb23af1",["D",1780052949625],"// Shared renderer preamble: React import, h, Icon, normalizeData, format helpers.\n// Consumed by frontend_renderers.ts + frontend_renderers2.ts (no build step on val.town).\nexport const PREAMBLE = `\n\nimport React from 'react';\nconst h = React.createElement;\nconst _ui = await import('/api/app-ui');\nconst { Icon } = _ui;\n\n// --- DATA NORMALIZATION -----------------------------------------------------\n// For integrations without their own custom renderer, flatten nested API responses\n// into rows the generic DataDisplay can render as a table.\nexport function normalizeData(name, endpoint, data) {\n if (!data || typeof data !== 'object') return data;\n switch (name) {\n case 'gsc':\n if (Array.isArray(data.rows)) return data.rows.map(r => ({\n [endpoint === 'pages' ? 'page' : 'query']: (r.keys || [])[0] || '',\n clicks: r.clicks, impressions: r.impressions,\n ctr: r.ctr != null ? (r.ctr * 100).toFixed(2) + '%' : '-',\n position: r.position != null ? r.position.toFixed(1) : '-'\n }));\n if (Array.isArray(data.sitemap)) return data.sitemap.map(s => ({\n path: s.path, type: s.type, lastSubmitted: s.lastSubmitted?.slice(0,10),\n isPending: s.isPending, warnings: s.warnings, errors: s.errors, contents: s.contents?.length\n }));\n return data;\n case 'youtube':\n if (Array.isArray(data.items)) return data.items.map(item => ({\n id: item.id?.videoId || item.id?.channelId || (typeof item.id === 'string' ? item.id : ''),\n title: item.snippet?.title || '', channel: item.snippet?.channelTitle || '',\n published: item.snippet?.publishedAt?.slice(0,10) || '',\n views: item.statistics?.viewCount || '', likes: item.statistics?.likeCount || '',\n comments: item.statistics?.commentCount || '', subscribers: item.statistics?.subscriberCount || ''\n }));\n return data;\n case 'producthunt': {\n const edges = (data?.data?.posts?.edges?.length ? data.data.posts.edges : data?.data?.topics?.edges) || [];\n if (edges.length) return edges.map(e => {\n const n = e.node || e;\n return { id:n.id, name:n.name, tagline:n.tagline||'', slug:n.slug||'', votes:n.votesCount||'', followers:n.followersCount||'', comments:n.commentsCount||'', url:n.url||'' };\n });\n return data;\n }\n case 'wordpress':\n if (Array.isArray(data)) return data.map(p => ({\n id:p.id, date:p.date?.slice(0,10)||'', status:p.status||'',\n title:p.title?.rendered||p.title||'', link:p.link||'', type:p.type||'',\n author:p.author, commentCount:p.comment_count\n }));\n return data;\n case 'outlook':\n if (Array.isArray(data.value)) return data.value.map(m => ({\n subject: m.subject || '(no subject)',\n from: m.from?.emailAddress?.name || m.from?.emailAddress?.address || '',\n received: m.receivedDateTime?.slice(0,16) || m.start?.dateTime?.slice(0,16) || '',\n isRead: m.isRead != null ? (m.isRead ? 'Yes' : 'No') : '',\n location: m.location?.displayName || '',\n end: m.end?.dateTime?.slice(0,16) || ''\n }));\n return data;\n case 'linkedin':\n if (Array.isArray(data.elements)) return data.elements.map(e => ({\n id: e.id || '', urn: e.ugcPost || e.author || '',\n created: e.created?.time ? new Date(e.created.time).toISOString().slice(0,10) : '',\n lifecycle: e.lifecycleState || ''\n }));\n if (data.localizedFirstName || data.firstName) {\n const first = data.localizedFirstName || Object.values(data.firstName?.localized || {})[0] || '';\n const last = data.localizedLastName || Object.values(data.lastName?.localized || {})[0] || '';\n return { name: (first + ' ' + last).trim(), id: data.id, headline: data.localizedHeadline || '' };\n }\n return data;\n default: return data;\n }\n}\n\n// --- HELPERS -----------------------------------------------------------------\nconst fmtRelative = (iso) => {\n if (!iso) return '';\n const ms = Date.now() - new Date(iso).getTime();\n if (ms < 60000) return 'just now';\n if (ms < 3600000) return Math.floor(ms/60000) + 'm ago';\n if (ms < 86400000) return Math.floor(ms/3600000) + 'h ago';\n if (ms < 604800000) return Math.floor(ms/86400000) + 'd ago';\n return new Date(iso).toLocaleDateString('en-US', {month:'short', day:'numeric'});\n};\nconst fmtFileSize = (bytes) => {\n const n = parseInt(bytes,10);\n if (!n) return '-';\n if (n < 1024) return n + ' B';\n if (n < 1048576) return (n/1024).toFixed(1) + ' KB';\n if (n < 1073741824) return (n/1048576).toFixed(1) + ' MB';\n return (n/1073741824).toFixed(2) + ' GB';\n};\nconst initialsOf = (name) => (name||'?').replace(/<[^>]+>/g,'').trim().split(/\\\\s+/).slice(0,2).map(s=>s[0]||'').join('').toUpperCase() || '?';\nconst colorFromName = (name) => {\n const colors = ['#c0c1ff','#4edea3','#d0bcff','#8083ff','#6ffbbe','#a078ff','#e1e0ff','#494bd6'];\n const sum = (name||'').split('').reduce((a,c) => a + c.charCodeAt(0), 0);\n return colors[sum % colors.length];\n};\n`;\n",[],[359,696],["D",1780052949625],[425,426,427,428,429,430,431,432,433,434,435,436,437,438,439,440,441],[359],"5d56fa0c-5a6b-11f1-bd26-ee650bb23af1","tasks.ts","97a41542-589e-11f1-b2d9-ee650bb23af1",434,["D",1779955259903],"// Tasks — Phase 1 Framework Layer.\n// Lightweight task tracking that can inherit ICE scoring from a parent\n// experiment, and that the LLM can populate from a KR gap or experiment\n// hypothesis so users don't have to draft work themselves.\nimport { sqlite } from \"../db.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\nimport { propagateTaskDone } from \"./propagate.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport interface Task {\n id: string;\n title: string;\n description: string | null;\n status: string; // todo | doing | done | blocked\n priority: string; // low | medium | high | urgent\n workspace: string | null;\n assignee: string | null;\n due_date: string | null;\n objective_id: string | null;\n key_result_id: string | null;\n experiment_id: string | null;\n ice_score: number | null;\n impact: number | null;\n confidence: number | null;\n ease: number | null;\n tags: string[];\n created_at: string;\n updated_at: string;\n completed_at: string | null;\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp10 = (n: any) => {\n const v = Number(n);\n if (!Number.isFinite(v)) return null;\n return Math.max(1, Math.min(10, Math.round(v)));\n};\n\nfunction parseTags(raw: any): string[] {\n if (!raw) return [];\n if (Array.isArray(raw)) return raw.map(String);\n try {\n const parsed = JSON.parse(String(raw));\n return Array.isArray(parsed) ? parsed.map(String) : [];\n } catch {\n return [];\n }\n}\n\nfunction rowToTask(columns: string[], row: any[]): Task {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n status: String(obj.status ?? \"todo\"),\n priority: String(obj.priority ?? \"medium\"),\n workspace: obj.workspace ?? null,\n assignee: obj.assignee ?? null,\n due_date: obj.due_date ?? null,\n objective_id: obj.objective_id ?? null,\n key_result_id: obj.key_result_id ?? null,\n experiment_id: obj.experiment_id ?? null,\n ice_score: obj.ice_score != null ? Number(obj.ice_score) : null,\n impact: obj.impact != null ? Number(obj.impact) : null,\n confidence: obj.confidence != null ? Number(obj.confidence) : null,\n ease: obj.ease != null ? Number(obj.ease) : null,\n tags: parseTags(obj.tags),\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n completed_at: obj.completed_at ?? null,\n };\n}\n\nconst TASK_COLS = `id, title, description, status, priority, workspace, assignee,\n due_date, objective_id, key_result_id, experiment_id, ice_score,\n impact, confidence, ease, tags, created_at, updated_at, completed_at`;\n\nasync function getTaskById(id: string): Promise<Task | null> {\n const r = await sqlite.execute(\n `SELECT ${TASK_COLS} FROM tasks WHERE id = ?`,\n [id],\n );\n if (!r.rows?.length) return null;\n return rowToTask(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listTasks(filters: {\n workspace?: string;\n status?: string;\n assignee?: string;\n objective_id?: string;\n experiment_id?: string;\n} = {}): Promise<Task[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.workspace) { where.push(\"workspace = ?\"); args.push(filters.workspace); }\n if (filters.status) { where.push(\"status = ?\"); args.push(filters.status); }\n if (filters.assignee) { where.push(\"assignee = ?\"); args.push(filters.assignee); }\n if (filters.objective_id) { where.push(\"objective_id = ?\"); args.push(filters.objective_id); }\n if (filters.experiment_id){ where.push(\"experiment_id = ?\");args.push(filters.experiment_id); }\n // ORDER BY: due_date ASC NULLS LAST, then created_at DESC.\n // SQLite sorts NULLs first by default, so we use a CASE expression.\n const sql = `SELECT ${TASK_COLS}\n FROM tasks\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY (CASE WHEN due_date IS NULL THEN 1 ELSE 0 END) ASC,\n due_date ASC,\n created_at DESC`;\n const r = await sqlite.execute(sql, args);\n return (r.rows || []).map((row) => rowToTask(r.columns, row as any[]));\n } catch {\n return [];\n }\n}\n\nexport async function listIceRankedTasks(limit = 20): Promise<Task[]> {\n try {\n const r = await sqlite.execute(\n `SELECT ${TASK_COLS}\n FROM tasks\n WHERE ice_score IS NOT NULL\n ORDER BY ice_score DESC\n LIMIT ?`,\n [limit],\n );\n return (r.rows || []).map((row) => rowToTask(r.columns, row as any[]));\n } catch {\n return [];\n }\n}\n\n// ─── Write ───────────────────────────────────────────────────────────────────\n\nexport async function createTask(data: Partial<Task>): Promise<Task> {\n // NOTE: tasks.id is INTEGER AUTOINCREMENT (db.ts) — do NOT supply a UUID here\n // (that caused SQLITE_MISMATCH and broke all task creation). Let SQLite assign it.\n const now = new Date().toISOString();\n\n let impact = data.impact ?? null;\n let confidence = data.confidence ?? null;\n let ease = data.ease ?? null;\n let ice_score = data.ice_score ?? null;\n\n // Inherit ICE from the parent experiment when one is linked + caller didn't\n // override. Lets users say \"make a task for this experiment\" without losing\n // the experiment's prioritization context.\n if (data.experiment_id && (impact == null || confidence == null || ease == null || ice_score == null)) {\n try {\n const exp = await sqlite.execute(\n \"SELECT impact, confidence, ease FROM experiments WHERE id = ?\",\n [data.experiment_id],\n );\n if (exp.rows?.length) {\n const row = exp.rows[0] as any[];\n const expI = Number(row[0]); const expC = Number(row[1]); const expE = Number(row[2]);\n if (impact == null) impact = expI;\n if (confidence == null) confidence = expC;\n if (ease == null) ease = expE;\n if (ice_score == null && Number.isFinite(expI) && Number.isFinite(expC) && Number.isFinite(expE)) {\n ice_score = Math.round(((expI + expC + expE) / 3) * 100) / 100;\n }\n }\n } catch { /* non-critical */ }\n }\n\n // If impact/confidence/ease are all set but ice_score isn't, compute it.\n if (ice_score == null && impact != null && confidence != null && ease != null) {\n const i = clamp10(impact); const c = clamp10(confidence); const e = clamp10(ease);\n if (i !== null && c !== null && e !== null) {\n ice_score = Math.round(((i + c + e) / 3) * 100) / 100;\n }\n }\n\n const tags = JSON.stringify(Array.isArray(data.tags) ? data.tags : []);\n const status = data.status ?? \"todo\";\n const completed_at = status === \"done\" ? now : null;\n\n await sqlite.execute(\n `INSERT INTO tasks (title, description, status, priority, workspace, assignee,\n due_date, objective_id, key_result_id, experiment_id, ice_score,\n impact, confidence, ease, tags, created_at, updated_at, completed_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled Task\",\n data.description ?? null,\n status,\n data.priority ?? \"medium\",\n data.workspace ?? null,\n data.assignee ?? null,\n data.due_date ?? null,\n data.objective_id ?? null,\n data.key_result_id ?? null,\n data.experiment_id ?? null,\n ice_score,\n impact, confidence, ease,\n tags,\n now, now, completed_at,\n ],\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((idRow.rows[0] as any[])[0]);\n return (await getTaskById(id))!;\n}\n\nexport async function updateTask(\n id: string,\n data: Partial<Task>,\n): Promise<Task | null> {\n try {\n const cur = await getTaskById(id);\n if (!cur) return null;\n const now = new Date().toISOString();\n\n const nextStatus = data.status ?? cur.status;\n // Transition to 'done' stamps completed_at; reverting clears it.\n const completed_at = (() => {\n if (nextStatus === \"done\" && cur.status !== \"done\") return now;\n if (nextStatus !== \"done\" && cur.status === \"done\") return null;\n return cur.completed_at;\n })();\n\n const next = {\n title: data.title ?? cur.title,\n description: \"description\" in data ? (data.description ?? null) : cur.description,\n status: nextStatus,\n priority: data.priority ?? cur.priority,\n workspace: \"workspace\" in data ? (data.workspace ?? null) : cur.workspace,\n assignee: \"assignee\" in data ? (data.assignee ?? null) : cur.assignee,\n due_date: \"due_date\" in data ? (data.due_date ?? null) : cur.due_date,\n objective_id: \"objective_id\" in data ? (data.objective_id ?? null) : cur.objective_id,\n key_result_id: \"key_result_id\" in data ? (data.key_result_id ?? null) : cur.key_result_id,\n experiment_id: \"experiment_id\" in data ? (data.experiment_id ?? null) : cur.experiment_id,\n impact: data.impact != null ? clamp10(data.impact) : cur.impact,\n confidence: data.confidence != null ? clamp10(data.confidence) : cur.confidence,\n ease: data.ease != null ? clamp10(data.ease) : cur.ease,\n tags: \"tags\" in data\n ? JSON.stringify(Array.isArray(data.tags) ? data.tags : [])\n : JSON.stringify(cur.tags),\n };\n\n // Recompute ICE if any of the three components are now present.\n let ice_score = cur.ice_score;\n if (next.impact != null && next.confidence != null && next.ease != null) {\n ice_score = Math.round(((next.impact + next.confidence + next.ease) / 3) * 100) / 100;\n } else if (\"ice_score\" in data) {\n ice_score = data.ice_score ?? null;\n }\n\n await sqlite.execute(\n `UPDATE tasks SET\n title = ?, description = ?, status = ?, priority = ?, workspace = ?,\n assignee = ?, due_date = ?, objective_id = ?, key_result_id = ?,\n experiment_id = ?, ice_score = ?, impact = ?, confidence = ?, ease = ?,\n tags = ?, updated_at = ?, completed_at = ?\n WHERE id = ?`,\n [\n next.title, next.description, next.status, next.priority, next.workspace,\n next.assignee, next.due_date, next.objective_id, next.key_result_id,\n next.experiment_id, ice_score, next.impact, next.confidence, next.ease,\n next.tags, now, completed_at, id,\n ],\n );\n // Closed-loop: a task moving INTO 'done' propagates to the measure layer\n // (executed-work counter + calendar marker + objective progress refresh).\n if (nextStatus === \"done\" && cur.status !== \"done\") {\n await propagateTaskDone({\n title: next.title,\n workspace: next.workspace,\n objective_id: next.objective_id,\n key_result_id: next.key_result_id,\n });\n }\n return await getTaskById(id);\n } catch {\n return null;\n }\n}\n\nexport async function deleteTask(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM tasks WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── LLM-generated task list ────────────────────────────────────────────────\n\ninterface GeneratedTask {\n title: string;\n description?: string;\n priority?: string;\n impact?: number;\n confidence?: number;\n ease?: number;\n}\n\nfunction parseTasksJson(raw: string): GeneratedTask[] {\n const tryParse = (s: string): GeneratedTask[] | null => {\n try {\n const parsed = JSON.parse(s);\n if (Array.isArray(parsed)) return parsed as GeneratedTask[];\n if (parsed && Array.isArray(parsed.tasks)) return parsed.tasks as GeneratedTask[];\n return null;\n } catch { return null; }\n };\n const direct = tryParse(raw);\n if (direct) return direct;\n // Regex-fenced fallback: pull the first JSON array out of mixed text.\n const arrMatch = raw.match(/\\[[\\s\\S]*\\]/);\n if (arrMatch) {\n const parsed = tryParse(arrMatch[0]);\n if (parsed) return parsed;\n }\n const objMatch = raw.match(/\\{[\\s\\S]*\\}/);\n if (objMatch) {\n const parsed = tryParse(objMatch[0]);\n if (parsed) return parsed;\n }\n return [];\n}\n\nexport async function generateTasksFromSource(input: {\n source_type: \"experiment\" | \"key_result\" | \"objective\";\n source_id: string;\n}): Promise<Task[]> {\n try {\n const config = await getLLMConfig();\n if (!config.api_key) return [];\n\n // Pull source context so the LLM has something concrete to plan against.\n let contextBlock = \"\";\n let link: Partial<Task> = {};\n if (input.source_type === \"experiment\") {\n const r = await sqlite.execute(\n `SELECT title, hypothesis, funnel_stage, aarrr_lever, success_metric, baseline_value\n FROM experiments WHERE id = ?`,\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Experiment: ${row[0]}\\n` +\n `Hypothesis: ${row[1] ?? \"(none)\"}\\n` +\n `Funnel stage: ${row[2] ?? \"(unspecified)\"}\\n` +\n `AARRR lever: ${row[3] ?? \"(unspecified)\"}\\n` +\n `Success metric: ${row[4] ?? \"(unspecified)\"}\\n` +\n `Baseline: ${row[5] ?? \"(unset)\"}`;\n link = { experiment_id: input.source_id };\n } else if (input.source_type === \"key_result\") {\n const r = await sqlite.execute(\n `SELECT title, metric_source, metric_key, start_value, current_value, target_value,\n unit, objective_id\n FROM key_results WHERE id = ?`,\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Key Result: ${row[0]}\\n` +\n `Metric: ${row[1] ?? \"?\"}:${row[2] ?? \"?\"}\\n` +\n `Progress: ${row[4]} / ${row[5]} ${row[6] ?? \"\"} (started at ${row[3]})\\n` +\n `Gap to close: ${Number(row[5] ?? 0) - Number(row[4] ?? 0)} ${row[6] ?? \"\"}`;\n link = { key_result_id: input.source_id, objective_id: row[7] ?? null };\n } else {\n const r = await sqlite.execute(\n \"SELECT title, description, quarter FROM objectives WHERE id = ?\",\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Objective: ${row[0]}\\n` +\n `Description: ${row[1] ?? \"(none)\"}\\n` +\n `Quarter: ${row[2] ?? \"(unspecified)\"}`;\n link = { objective_id: input.source_id };\n }\n\n const system =\n \"You are an operations planner breaking work into concrete tasks. \" +\n \"Return ONLY a valid JSON array (up to 5 items) of \" +\n '{\"title\":string,\"description\":string,\"priority\":\"low|medium|high|urgent\",' +\n '\"impact\":1-10,\"confidence\":1-10,\"ease\":1-10}. ' +\n \"No markdown fences, no prose outside the array.\";\n\n const user =\n `Plan up to 5 concrete next-step tasks for the following:\\n\\n${contextBlock}\\n\\n` +\n `Each task should be one focused action a single person can do in <1 week.`;\n\n let raw = \"\";\n try {\n const res = await callLLM(\n config,\n [{ role: \"user\", content: user } as any],\n { systemPrompt: system, maxTokens: 800, temperature: 0.4 },\n );\n raw = res.content || \"\";\n } catch {\n return [];\n }\n\n const generated = parseTasksJson(raw).slice(0, 5);\n if (!generated.length) return [];\n\n const inserted: Task[] = [];\n for (const g of generated) {\n if (!g.title) continue;\n const created = await createTask({\n title: String(g.title).slice(0, 200),\n description: g.description ? String(g.description).slice(0, 1000) : null,\n priority: [\"low\", \"medium\", \"high\", \"urgent\"].includes(String(g.priority))\n ? String(g.priority)\n : \"medium\",\n status: \"todo\",\n impact: g.impact != null ? clamp10(g.impact) ?? undefined : undefined,\n confidence: g.confidence != null ? clamp10(g.confidence) ?? undefined : undefined,\n ease: g.ease != null ? clamp10(g.ease) ?? undefined : undefined,\n ...link,\n } as Partial<Task>);\n inserted.push(created);\n }\n return inserted;\n } catch {\n return [];\n }\n}\n",[],[365,706],"c9f08122-5d65-11f1-8d2a-1607ee4eb77e","stats.ts","c9f08208-5d65-11f1-8d2a-1607ee4eb77e",602,["D",1780282718615],"// Part 3 data-spine — real statistics over metric_history (replaces the fabricated\n// constant-confidence correlation rules in intelligence.getCorrelations and the\n// \"typed a bigger number = win\" experiment outcomes in experiments.completeExperiment).\n//\n// Pure, dependency-free numeric helpers. Everything here is honest: if there is not\n// enough data, callers get null/insufficient — never a made-up coefficient.\n\nexport interface SeriesPoint { date: string; value: number }\n\nfunction mean(xs: number[]): number {\n return xs.length ? xs.reduce((a, b) => a + b, 0) / xs.length : 0;\n}\n\nfunction std(xs: number[], m: number): number {\n if (xs.length < 2) return 0;\n const v = xs.reduce((a, b) => a + (b - m) * (b - m), 0) / (xs.length - 1);\n return Math.sqrt(v);\n}\n\n// Pearson product-moment correlation. Returns r in [-1, 1], or 0 if undefined\n// (constant series / fewer than 2 points).\nexport function pearson(xs: number[], ys: number[]): number {\n const n = Math.min(xs.length, ys.length);\n if (n < 2) return 0;\n const mx = mean(xs.slice(0, n));\n const my = mean(ys.slice(0, n));\n let num = 0, dx = 0, dy = 0;\n for (let i = 0; i < n; i++) {\n const a = xs[i] - mx, b = ys[i] - my;\n num += a * b; dx += a * a; dy += b * b;\n }\n if (dx === 0 || dy === 0) return 0;\n const r = num / Math.sqrt(dx * dy);\n return Math.max(-1, Math.min(1, r));\n}\n\n// Spearman rank correlation — robust to non-linear monotonic relationships.\nexport function spearman(xs: number[], ys: number[]): number {\n const n = Math.min(xs.length, ys.length);\n if (n < 2) return 0;\n const rank = (arr: number[]): number[] => {\n const idx = arr.map((v, i) => [v, i]).sort((a, b) => a[0] - b[0]);\n const r = new Array(arr.length).fill(0);\n for (let i = 0; i < idx.length;) {\n let j = i;\n while (j + 1 < idx.length && idx[j + 1][0] === idx[i][0]) j++;\n const avg = (i + j) / 2 + 1; // average rank for ties (1-based)\n for (let k = i; k <= j; k++) r[idx[k][1]] = avg;\n i = j + 1;\n }\n return r;\n };\n return pearson(rank(xs.slice(0, n)), rank(ys.slice(0, n)));\n}\n\n// Two-tailed p-value for a Pearson r over n samples, via the t-distribution\n// t = r * sqrt((n-2)/(1-r^2)) with df = n-2. Uses an incomplete-beta approximation\n// of the Student-t survival function (good enough for a significance gate, not a paper).\nexport function pValueFromPearson(r: number, n: number): number {\n if (n < 3) return 1;\n const rr = Math.min(0.999999, Math.abs(r));\n const df = n - 2;\n const t = rr * Math.sqrt(df / (1 - rr * rr));\n return studentTTwoTailed(t, df);\n}\n\n// Student-t two-tailed p-value via the regularized incomplete beta function.\nfunction studentTTwoTailed(t: number, df: number): number {\n const x = df / (df + t * t);\n // P(|T| > t) = I_x(df/2, 1/2)\n const p = betai(df / 2, 0.5, x);\n return Math.max(0, Math.min(1, p));\n}\n\n// Regularized incomplete beta I_x(a,b) via continued fraction (Numerical Recipes).\nfunction betai(a: number, b: number, x: number): number {\n if (x <= 0) return 0;\n if (x >= 1) return 1;\n const lbeta = gammaln(a + b) - gammaln(a) - gammaln(b);\n const front = Math.exp(lbeta + a * Math.log(x) + b * Math.log(1 - x));\n if (x < (a + 1) / (a + b + 2)) return front * betacf(a, b, x) / a;\n return 1 - front * betacf(b, a, 1 - x) / b;\n}\n\nfunction betacf(a: number, b: number, x: number): number {\n const MAXIT = 200, EPS = 3e-12, FPMIN = 1e-300;\n let qab = a + b, qap = a + 1, qam = a - 1;\n let c = 1, d = 1 - qab * x / qap;\n if (Math.abs(d) < FPMIN) d = FPMIN;\n d = 1 / d; let h = d;\n for (let m = 1; m <= MAXIT; m++) {\n const m2 = 2 * m;\n let aa = m * (b - m) * x / ((qam + m2) * (a + m2));\n d = 1 + aa * d; if (Math.abs(d) < FPMIN) d = FPMIN;\n c = 1 + aa / c; if (Math.abs(c) < FPMIN) c = FPMIN;\n d = 1 / d; h *= d * c;\n aa = -(a + m) * (qab + m) * x / ((a + m2) * (qap + m2));\n d = 1 + aa * d; if (Math.abs(d) < FPMIN) d = FPMIN;\n c = 1 + aa / c; if (Math.abs(c) < FPMIN) c = FPMIN;\n d = 1 / d; const del = d * c; h *= del;\n if (Math.abs(del - 1) < EPS) break;\n }\n return h;\n}\n\nfunction gammaln(xx: number): number {\n const cof = [76.18009172947146, -86.50532032941677, 24.01409824083091,\n -1.231739572450155, 0.1208650973866179e-2, -0.5395239384953e-5];\n let x = xx, y = xx;\n let tmp = x + 5.5;\n tmp -= (x + 0.5) * Math.log(tmp);\n let ser = 1.000000000190015;\n for (let j = 0; j < 6; j++) { y += 1; ser += cof[j] / y; }\n return -tmp + Math.log(2.5066282746310005 * ser / x);\n}\n\n// Align two daily series by date (inner join on the date string) → parallel value arrays.\nexport function alignSeries(a: SeriesPoint[], b: SeriesPoint[]): { xs: number[]; ys: number[] } {\n const mb = new Map(b.map((p) => [p.date, p.value]));\n const xs: number[] = [], ys: number[] = [];\n for (const p of a) {\n if (mb.has(p.date)) { xs.push(p.value); ys.push(mb.get(p.date) as number); }\n }\n return { xs, ys };\n}\n\n// Best lagged correlation: tries lags 0..maxLag where series B trails series A by `lag`\n// days (i.e. a change in A precedes a change in B). Returns the lag with the largest |r|,\n// the coefficient, its p-value, and the paired-sample count n at that lag.\nexport function bestLaggedCorrelation(\n a: SeriesPoint[],\n b: SeriesPoint[],\n maxLag = 7,\n method: \"pearson\" | \"spearman\" = \"pearson\",\n): { r: number; lag: number; n: number; p: number } {\n let best = { r: 0, lag: 0, n: 0, p: 1 };\n const shiftDate = (d: string, days: number): string => {\n const t = new Date(d + \"T00:00:00Z\").getTime() + days * 86400000;\n return new Date(t).toISOString().slice(0, 10);\n };\n for (let lag = 0; lag <= maxLag; lag++) {\n // Shift B back by `lag` so A[date] aligns with B[date+lag].\n const bShift = b.map((p) => ({ date: shiftDate(p.date, -lag), value: p.value }));\n const { xs, ys } = alignSeries(a, bShift);\n if (xs.length < 3) continue;\n const r = method === \"spearman\" ? spearman(xs, ys) : pearson(xs, ys);\n if (Math.abs(r) > Math.abs(best.r)) {\n best = { r, lag, n: xs.length, p: pValueFromPearson(r, xs.length) };\n }\n }\n return best;\n}\n\n// Two-proportion z-test for experiment significance (control vs variant conversion rates).\n// Returns z, the two-tailed p-value, and the pooled rates. Honest: needs both arms' counts.\nexport function twoProportionZTest(\n c1: number, n1: number, c2: number, n2: number,\n): { z: number; p: number; rate1: number; rate2: number; n: number } | null {\n if (n1 <= 0 || n2 <= 0) return null;\n const p1 = c1 / n1, p2 = c2 / n2;\n const pPool = (c1 + c2) / (n1 + n2);\n const se = Math.sqrt(pPool * (1 - pPool) * (1 / n1 + 1 / n2));\n if (se === 0) return null;\n const z = (p2 - p1) / se;\n const p = 2 * (1 - normalCdf(Math.abs(z)));\n return { z, p: Math.max(0, Math.min(1, p)), rate1: p1, rate2: p2, n: n1 + n2 };\n}\n\n// Standard normal CDF (Abramowitz & Stegun 7.1.26 erf approximation).\nexport function normalCdf(x: number): number {\n const t = 1 / (1 + 0.2316419 * Math.abs(x));\n const d = 0.3989422804014327 * Math.exp(-x * x / 2);\n const prob = d * t * (0.319381530 + t * (-0.356563782 + t * (1.781477937 + t * (-1.821255978 + t * 1.330274429))));\n return x > 0 ? 1 - prob : prob;\n}\n\n// Minimum paired samples before a correlation is allowed to surface, and the |r| floor.\nexport const MIN_CORR_SAMPLES = 8;\nexport const MIN_CORR_ABS_R = 0.5;\nexport const MAX_CORR_P = 0.1;\n\n// Classify |r| into the app's strength buckets.\nexport function corrStrength(absR: number): \"weak\" | \"moderate\" | \"strong\" {\n if (absR >= 0.7) return \"strong\";\n if (absR >= 0.5) return \"moderate\";\n return \"weak\";\n}\n",[],[365,714],"ea50738e-5d6d-11f1-95b0-1607ee4eb77e","skills.ts","ea507456-5d6d-11f1-95b0-1607ee4eb77e",643,["D",1780286208905],"// Part 1 — Skills service. Mirrors knowledge.ts style. Bodies come from the bundled\n// catalog (skills/index.ts, read from memory); the `skills` DB table holds only per-skill\n// STATE (enabled/pinned) + user-created custom skills. getEffectiveSkills() merges them.\nimport { sqlite } from \"../db.ts\";\nimport {\n BUILTIN_SKILLS, getSkillDef, getSkillBody, listReferences, getReference,\n type SkillDef,\n} from \"../skills/index.ts\";\nimport { getCompanyContextString } from \"./company.ts\";\n\nconst APPLY_CAP = 12000; // max chars of a skill body returned by applySkillContext\nconst INDEX_CAP = 6500; // max chars of the prompt-injected skills index\n\nexport interface EffectiveSkill {\n name: string;\n category: string;\n description: string;\n version: string;\n enabled: boolean;\n pinned: boolean;\n source: \"builtin\" | \"user\";\n has_body: boolean;\n}\n\ninterface SkillStateRow { enabled: number; pinned: number; source: string; category: string | null; description: string | null; body: string | null }\n\nasync function loadState(): Promise<Map<string, SkillStateRow>> {\n const map = new Map<string, SkillStateRow>();\n try {\n const r = await sqlite.execute(\"SELECT name, enabled, pinned, source, category, description, body FROM skills\");\n for (const row of (r.rows || []) as any[]) {\n const name = String(row[0]);\n map.set(name, {\n enabled: Number(row[1]), pinned: Number(row[2]), source: String(row[3] || \"builtin\"),\n category: row[4] != null ? String(row[4]) : null,\n description: row[5] != null ? String(row[5]) : null,\n body: row[6] != null ? String(row[6]) : null,\n });\n }\n } catch { /* table missing → all builtins default-enabled */ }\n return map;\n}\n\n// Merge builtin catalog + DB state + custom (source='user') skills.\nexport async function getEffectiveSkills(): Promise<EffectiveSkill[]> {\n const state = await loadState();\n const out: EffectiveSkill[] = [];\n const seen = new Set<string>();\n for (const b of BUILTIN_SKILLS) {\n const s = state.get(b.name);\n out.push({\n name: b.name, category: b.category, description: b.description, version: b.version,\n enabled: s ? s.enabled !== 0 : true,\n pinned: s ? s.pinned === 1 : false,\n source: \"builtin\", has_body: true,\n });\n seen.add(b.name);\n }\n for (const [name, s] of state) {\n if (seen.has(name) || s.source !== \"user\") continue;\n out.push({\n name, category: s.category || \"Custom\", description: s.description || \"\", version: \"1.0.0\",\n enabled: s.enabled !== 0, pinned: s.pinned === 1, source: \"user\", has_body: !!s.body,\n });\n }\n return out;\n}\n\n// Grouped-by-category list (no body) for the Playbooks page + GET /api/skills.\nexport async function listSkills(): Promise<{\n categories: Array<{ category: string; skills: EffectiveSkill[] }>;\n total: number; enabled: number;\n}> {\n const all = await getEffectiveSkills();\n const byCat = new Map<string, EffectiveSkill[]>();\n for (const s of all) {\n if (!byCat.has(s.category)) byCat.set(s.category, []);\n byCat.get(s.category)!.push(s);\n }\n const categories = [...byCat.entries()]\n .map(([category, skills]) => ({ category, skills: skills.sort((a, b) => a.name.localeCompare(b.name)) }))\n .sort((a, b) => a.category.localeCompare(b.category));\n return { categories, total: all.length, enabled: all.filter((s) => s.enabled).length };\n}\n\n// Full skill (body + reference list). Custom bodies come from the DB; builtin from catalog.\nexport async function getSkill(name: string): Promise<\n (EffectiveSkill & { body: string; references: string[] }) | null\n> {\n const eff = (await getEffectiveSkills()).find((s) => s.name === name);\n if (!eff) return null;\n let body = getSkillBody(name) ?? \"\";\n let refs = listReferences(name);\n if (eff.source === \"user\") {\n const state = await loadState();\n body = state.get(name)?.body || \"\";\n refs = [];\n }\n return { ...eff, body, references: refs };\n}\n\n// Compact index of ENABLED skills for prompt injection: \"- name — first sentence [Category]\".\nexport async function getSkillsIndex(opts: { maxChars?: number } = {}): Promise<string> {\n const cap = opts.maxChars ?? INDEX_CAP;\n const all = (await getEffectiveSkills()).filter((s) => s.enabled);\n if (!all.length) return \"\";\n all.sort((a, b) => (b.pinned ? 1 : 0) - (a.pinned ? 1 : 0) || a.category.localeCompare(b.category) || a.name.localeCompare(b.name));\n const lines: string[] = [];\n let used = 0;\n for (const s of all) {\n const first = (s.description.split(/(?<=[.!?])\\s/)[0] || s.description).slice(0, 160).trim();\n const line = `- ${s.name} — ${first} [${s.category}]`;\n if (used + line.length + 1 > cap) break;\n lines.push(line); used += line.length + 1;\n }\n return \"MARKETING SKILLS INDEX (call apply_skill with the skill name when a request matches one):\\n\" + lines.join(\"\\n\");\n}\n\n// The string the apply_skill tool returns: brand/company context + the full playbook body.\nexport async function applySkillContext(name: string): Promise<string | null> {\n const def = getSkillDef(name);\n let body = def?.body ?? null;\n if (!body) {\n const state = await loadState();\n body = state.get(name)?.body ?? null;\n }\n if (!body) return null;\n let brand = \"\";\n try { brand = await getCompanyContextString(1500); } catch { /* optional */ }\n const head = brand ? brand + \"\\n\\n---\\n\\n\" : \"\";\n return (head + body).slice(0, APPLY_CAP + head.length);\n}\n\n// Reference doc fetch for the get_skill_reference tool.\nexport function getSkillReference(name: string, ref: string): string | null {\n return getReference(name, ref);\n}\nexport function getSkillReferenceList(name: string): string[] {\n return listReferences(name);\n}\n\nexport async function setSkillState(name: string, patch: { enabled?: boolean; pinned?: boolean }): Promise<boolean> {\n const isBuiltin = !!getSkillDef(name);\n const state = await loadState();\n const existing = state.get(name);\n if (!isBuiltin && !existing) return false; // unknown skill\n const cat = existing?.category ?? getSkillDef(name)?.category ?? null;\n const desc = existing?.description ?? getSkillDef(name)?.description ?? null;\n const enabled = patch.enabled != null ? (patch.enabled ? 1 : 0) : (existing ? existing.enabled : 1);\n const pinned = patch.pinned != null ? (patch.pinned ? 1 : 0) : (existing ? existing.pinned : 0);\n await sqlite.execute(\n `INSERT INTO skills (name, enabled, pinned, source, category, description, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, datetime('now'))\n ON CONFLICT(name) DO UPDATE SET enabled=excluded.enabled, pinned=excluded.pinned, updated_at=datetime('now')`,\n [name, enabled, pinned, existing?.source ?? (isBuiltin ? \"builtin\" : \"user\"), cat, desc],\n );\n return true;\n}\n\nexport async function createCustomSkill(data: { name: string; category?: string; description?: string; body: string }): Promise<{ ok: boolean; error?: string }> {\n const name = String(data.name || \"\").trim().toLowerCase().replace(/\\s+/g, \"-\");\n if (!name) return { ok: false, error: \"name required\" };\n if (getSkillDef(name)) return { ok: false, error: \"name collides with a builtin skill\" };\n if (!data.body || data.body.length < 10) return { ok: false, error: \"body too short\" };\n if (data.body.length > 60000) return { ok: false, error: \"body too large (cap 60KB)\" };\n await sqlite.execute(\n `INSERT INTO skills (name, enabled, pinned, source, category, description, body, updated_at)\n VALUES (?, 1, 0, 'user', ?, ?, ?, datetime('now'))\n ON CONFLICT(name) DO UPDATE SET category=excluded.category, description=excluded.description, body=excluded.body, updated_at=datetime('now')`,\n [name, data.category || \"Custom\", (data.description || \"\").slice(0, 1000), data.body],\n );\n return { ok: true };\n}\n",[],[365,722],"2af1af84-5d69-11f1-b581-1607ee4eb77e","signals.ts","9808cfb4-589e-11f1-9c4f-ee650bb23af1",623,["D",1780284169846],"// Intent signals — captures and scores buying-intent events per lead so the team can\n// triage hot leads in real time. Signals are typed (email_reply, pricing_keyword, etc.)\n// and each carries both a strength (1-3) and a score (5-50). Hot leads = leads whose\n// 30-day unactioned signal score exceeds 50.\n//\n// Sourced opportunistically from Gmail (via intelligence.getLeadScores) so the table\n// stays fresh without requiring a separate webhook ingestion pipeline.\nimport { sqlite } from \"../db.ts\";\n\n// ─── Weights ─────────────────────────────────────────────────────────────────\n// strength: 1 (weak) / 2 (medium) / 3 (strong) — used by UI badge.\n// score: numeric contribution to hot-lead score (summed over 30d window).\nexport const SIGNAL_WEIGHTS: Record<string, { strength: number; score: number }> = {\n email_open: { strength: 1, score: 5 },\n email_reply: { strength: 3, score: 25 },\n pricing_keyword: { strength: 3, score: 30 },\n demo_request: { strength: 3, score: 40 },\n linkedin_view: { strength: 1, score: 10 },\n calendar_event: { strength: 3, score: 50 },\n content_download: { strength: 2, score: 20 },\n};\n\nexport interface IntentSignal {\n id: string;\n lead_email: string;\n lead_name: string | null;\n company: string | null;\n signal_type: string;\n signal_source: string;\n signal_strength: number;\n raw_data: any;\n detected_at: string;\n actioned: number;\n actioned_at: string | null;\n}\n\nfunction rowToSignal(columns: string[], row: any[]): IntentSignal {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n let raw: any = null;\n try { raw = obj.raw_data ? JSON.parse(String(obj.raw_data)) : null; } catch { raw = String(obj.raw_data ?? \"\"); }\n return {\n id: String(obj.id),\n lead_email: String(obj.lead_email),\n lead_name: obj.lead_name ?? null,\n company: obj.company ?? null,\n signal_type: String(obj.signal_type),\n signal_source: String(obj.signal_source ?? \"\"),\n signal_strength: Number(obj.signal_strength) || 1,\n raw_data: raw,\n detected_at: String(obj.detected_at),\n actioned: Number(obj.actioned) || 0,\n actioned_at: obj.actioned_at ?? null,\n };\n}\n\n// ─── CRUD ─────────────────────────────────────────────────────────────────────\n\nexport async function listSignals(filters?: { lead_email?: string; actioned?: boolean }): Promise<IntentSignal[]> {\n // Best-effort refresh first — never fails the read if integrations are down.\n try { await syncSignalsFromIntegrations(); } catch { /* non-critical */ }\n\n const where: string[] = [];\n const args: any[] = [];\n if (filters?.lead_email) { where.push(\"lead_email = ?\"); args.push(filters.lead_email); }\n if (typeof filters?.actioned === \"boolean\") { where.push(\"actioned = ?\"); args.push(filters.actioned ? 1 : 0); }\n const sql = `SELECT id, lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned, actioned_at\n FROM intent_signals\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY detected_at DESC LIMIT 100`;\n try {\n const r = await sqlite.execute(sql, args);\n return (r.rows || []).map((row) => rowToSignal(r.columns, row as any[]));\n } catch { return []; }\n}\n\nexport async function recordSignal(data: {\n lead_email: string;\n lead_name?: string;\n company?: string;\n signal_type: string;\n signal_source?: string;\n source?: string; // alias accepted for convenience\n raw_data?: any;\n score?: number; // ignored — strength is derived from SIGNAL_WEIGHTS\n workspace?: string; // ignored at signal level\n}): Promise<IntentSignal> {\n const weight = SIGNAL_WEIGHTS[data.signal_type] ?? { strength: 1, score: 5 };\n const rawJson = data.raw_data != null ? JSON.stringify(data.raw_data) : null;\n const now = new Date().toISOString();\n const signal_source = data.signal_source ?? data.source ?? \"manual\";\n await sqlite.execute(\n `INSERT INTO intent_signals (lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, 0)`,\n [data.lead_email, data.lead_name ?? null, data.company ?? null,\n data.signal_type, signal_source, weight.strength, rawJson, now]\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return {\n id, lead_email: data.lead_email, lead_name: data.lead_name ?? null, company: data.company ?? null,\n signal_type: data.signal_type, signal_source,\n signal_strength: weight.strength, raw_data: data.raw_data ?? null,\n detected_at: now, actioned: 0, actioned_at: null,\n };\n}\n\nexport async function actionSignal(id: string): Promise<IntentSignal | null> {\n const now = new Date().toISOString();\n await sqlite.execute(\n \"UPDATE intent_signals SET actioned = 1, actioned_at = ? WHERE id = ?\",\n [now, id]\n );\n const r = await sqlite.execute(\n `SELECT id, lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned, actioned_at\n FROM intent_signals WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) return null;\n return rowToSignal(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Hot leads aggregation ───────────────────────────────────────────────────\n// Group unactioned signals from the last 30 days by lead_email, sum weighted\n// scores per signal_type, return leads with total > 50 sorted desc.\n\nexport interface HotLead {\n email: string;\n name: string;\n company: string;\n score: number;\n top_signals: string[];\n signal_count: number;\n last_detected_at: string;\n}\n\nexport async function getHotLeads(): Promise<HotLead[]> {\n let rows: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT lead_email, lead_name, company, signal_type, detected_at\n FROM intent_signals\n WHERE actioned = 0\n AND detected_at >= datetime('now','-30 days')\n ORDER BY detected_at DESC`\n );\n rows = (r.rows || []) as any[];\n } catch { rows = []; }\n\n // Aggregate by email.\n type Agg = {\n email: string;\n name: string;\n company: string;\n score: number;\n type_counts: Map<string, number>;\n signal_count: number;\n last_detected_at: string;\n };\n const byEmail = new Map<string, Agg>();\n for (const row of rows) {\n const email = String(row[0]);\n const name = String(row[1] ?? email.split(\"@\")[0]);\n const company = String(row[2] ?? \"\");\n const type = String(row[3]);\n const detected = String(row[4]);\n const weight = SIGNAL_WEIGHTS[type]?.score ?? 5;\n let agg = byEmail.get(email);\n if (!agg) {\n agg = { email, name, company, score: 0, type_counts: new Map(), signal_count: 0, last_detected_at: detected };\n byEmail.set(email, agg);\n }\n agg.score += weight;\n agg.signal_count += 1;\n agg.type_counts.set(type, (agg.type_counts.get(type) || 0) + 1);\n if (detected > agg.last_detected_at) agg.last_detected_at = detected;\n }\n\n const hot: HotLead[] = [];\n for (const agg of byEmail.values()) {\n if (agg.score <= 50) continue;\n // top_signals: 3 most frequent signal types for this lead\n const top = Array.from(agg.type_counts.entries())\n .sort((a, b) => b[1] - a[1])\n .slice(0, 3)\n .map(([t]) => t);\n hot.push({\n email: agg.email,\n name: agg.name,\n company: agg.company,\n score: agg.score,\n top_signals: top,\n signal_count: agg.signal_count,\n last_detected_at: agg.last_detected_at,\n });\n }\n hot.sort((a, b) => b.score - a.score);\n return hot;\n}\n\nexport async function getHotLeadsCount(): Promise<number> {\n const leads = await getHotLeads();\n return leads.length;\n}\n\n// ─── Sync from intelligence layer ────────────────────────────────────────────\n// Pulls fresh lead scores from intelligence.getLeadScores and converts MQL/SQL\n// behavior into intent_signals rows. Idempotent-ish: skips signals that already\n// exist (same email + type within the last 24h) to avoid runaway duplicates.\n\nexport async function syncSignalsFromIntegrations(): Promise<{ recorded: number; skipped: number }> {\n let recorded = 0;\n let skipped = 0;\n try {\n const { getLeadScores } = await import(\"./intelligence.ts\");\n const leads = await getLeadScores();\n for (const lead of leads) {\n if ((lead.mql_score ?? 0) < 25) continue;\n // Determine which signal types to emit based on what the lead's breakdown shows.\n const wants: Array<{ type: string; source: string }> = [];\n const sigs = (lead.sql_signals || []) as string[];\n const bk = (lead.signal_breakdown || {}) as Record<string, number>;\n\n if (sigs.includes(\"reply_to_outreach_email\") || sigs.includes(\"replied_to_thread\")) {\n wants.push({ type: \"email_reply\", source: lead.source || \"gmail\" });\n }\n if (bk.visits_pricing || bk.gmail_pricing) {\n wants.push({ type: \"pricing_keyword\", source: lead.source || \"gmail\" });\n }\n if (sigs.includes(\"requested_demo_or_pricing\") || sigs.includes(\"requested_demo_or_trial\") || sigs.includes(\"requested_proposal\")) {\n wants.push({ type: \"demo_request\", source: lead.source || \"gmail\" });\n }\n if (bk.downloads_content) {\n wants.push({ type: \"content_download\", source: lead.source || \"gmail\" });\n }\n if (sigs.includes(\"booked_discovery_call\")) {\n wants.push({ type: \"calendar_event\", source: \"outlook\" });\n }\n if (wants.length === 0 && (lead.mql_score ?? 0) >= 40) {\n // High MQL score but no specific signal — log a generic email_open.\n wants.push({ type: \"email_open\", source: lead.source || \"gmail\" });\n }\n\n for (const w of wants) {\n // Dedup: skip if this email already has this signal_type within the last 24h.\n try {\n const dup = await sqlite.execute(\n `SELECT id FROM intent_signals\n WHERE lead_email = ? AND signal_type = ?\n AND detected_at >= datetime('now','-1 day') LIMIT 1`,\n [lead.email, w.type]\n );\n if (dup.rows?.length) { skipped++; continue; }\n } catch { /* tolerate query failure, attempt insert */ }\n\n try {\n await recordSignal({\n lead_email: lead.email,\n lead_name: lead.name,\n company: lead.company,\n signal_type: w.type,\n signal_source: w.source,\n raw_data: { mql_score: lead.mql_score, stage: lead.stage },\n });\n recorded++;\n } catch { /* swallow individual insert error */ }\n }\n }\n } catch {\n // intelligence module unavailable or threw — safe no-op.\n }\n return { recorded, skipped };\n}\n",[],[365,730],"2a9d71a8-5d69-11f1-a87a-1607ee4eb77e","propagate.ts","514dfcee-5a69-11f1-bcd7-ee650bb23af1",622,["D",1780284169295],"// Closed-loop propagation — when work is EXECUTED (a task is marked done or an\n// experiment is completed), reflect it in the MEASURE layer so execution and\n// measurement stop being decoupled:\n// 1) bump an internal \"executed work\" counter (metric_history, integration='internal')\n// 2) drop a marker onto the unified calendar (calendar_entries → 'custom' source)\n// 3) refresh the linked objective's progress (mean of its KR progress)\n//\n// Self-contained: imports ONLY sqlite to avoid import cycles with okrs/experiments/tasks.\nimport { sqlite } from \"../db.ts\";\n\nconst today = () => new Date().toISOString().slice(0, 10);\n\nasync function bumpExecutedWork(metricKey: string): Promise<void> {\n try {\n await sqlite.execute(\n \"INSERT INTO metric_history (integration, metric_key, value) VALUES ('internal', ?, 1)\",\n [metricKey],\n );\n } catch { /* non-critical */ }\n}\n\nasync function addCalendarMarker(title: string, workspace: string | null, color: string): Promise<void> {\n try {\n await sqlite.execute(\n `INSERT INTO calendar_entries (title, event_date, event_type, workspace, color, all_day)\n VALUES (?, ?, 'completion', ?, ?, 1)`,\n [title, today(), workspace ?? \"marketing\", color],\n );\n } catch { /* non-critical */ }\n}\n\n// Recompute one objective's progress = mean of its KR progress (matches okrs.computeObjectiveProgress).\nasync function refreshObjectiveProgress(objectiveId: string | number | null): Promise<void> {\n if (objectiveId == null) return;\n try {\n const r = await sqlite.execute(\n \"SELECT progress FROM key_results WHERE objective_id = ?\",\n [objectiveId],\n );\n const rows = (r.rows || []) as any[];\n if (!rows.length) return;\n const mean = Math.round(rows.reduce((a, row) => a + (Number(row[0]) || 0), 0) / rows.length);\n await sqlite.execute(\n \"UPDATE objectives SET progress = ?, updated_at = datetime('now') WHERE id = ?\",\n [mean, objectiveId],\n );\n } catch { /* non-critical */ }\n}\n\n// A task links to an objective directly or via its key_result.\nasync function objectiveForTask(t: { objective_id: any; key_result_id: any }): Promise<string | null> {\n if (t.objective_id != null) return String(t.objective_id);\n if (t.key_result_id != null) {\n try {\n const r = await sqlite.execute(\"SELECT objective_id FROM key_results WHERE id = ?\", [t.key_result_id]);\n const row = ((r.rows || [])[0]) as any[] | undefined;\n if (row && row[0] != null) return String(row[0]);\n } catch { /* */ }\n }\n return null;\n}\n\nexport async function propagateTaskDone(task: {\n title: string; workspace: string | null; objective_id: any; key_result_id: any;\n}): Promise<void> {\n await bumpExecutedWork(\"tasks_completed\");\n await addCalendarMarker(\"Done: \" + (task.title || \"task\"), task.workspace, \"green\");\n await refreshObjectiveProgress(await objectiveForTask(task));\n}\n\nexport async function propagateExperimentComplete(exp: {\n title: string; outcome: string | null; objective_id: any;\n}): Promise<void> {\n await bumpExecutedWork(\"experiments_completed\");\n if (exp.outcome === \"win\") await bumpExecutedWork(\"experiments_won\");\n await addCalendarMarker(\n \"Experiment \" + (exp.outcome || \"done\") + \": \" + (exp.title || \"\"),\n null,\n exp.outcome === \"win\" ? \"secondary\" : \"tertiary\",\n );\n await refreshObjectiveProgress(exp.objective_id != null ? String(exp.objective_id) : null);\n}\n\n// Executed-work counts over a window — read by intelligence for execution-aware KPIs.\nexport async function getExecutedWorkCounts(\n days = 30,\n): Promise<{ tasks_completed: number; experiments_completed: number; experiments_won: number }> {\n const out = { tasks_completed: 0, experiments_completed: 0, experiments_won: 0 };\n try {\n const r = await sqlite.execute(\n `SELECT metric_key, COUNT(*) FROM metric_history\n WHERE integration = 'internal' AND recorded_at >= datetime('now', ?)\n GROUP BY metric_key`,\n [`-${days} days`],\n );\n for (const row of (r.rows || []) as any[]) {\n const k = String(row[0]);\n if (k in out) (out as any)[k] = Number(row[1]) || 0;\n }\n } catch { /* */ }\n return out;\n}\n",[],[365,738],"2a2387bc-5d69-11f1-b2ca-1607ee4eb77e","okrs.ts","96e84dee-589e-11f1-8829-ee650bb23af1",621,["D",1780284168488],"// OKRs — Objectives + Key Results. Phase 1 Framework Layer.\n// Objectives roll up KR progress; KR current_value auto-syncs from the funnel\n// summary when metric_source is a connected integration.\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\nimport { getAttributionSummary } from \"./aio.ts\";\n\n// Current quarter as \"YYYY-Qn\" (runs on val.town where Date is available).\nfunction currentQuarter(): string {\n const d = new Date();\n return d.getFullYear() + \"-Q\" + (Math.floor(d.getMonth() / 3) + 1);\n}\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport interface KeyResult {\n id: string;\n objective_id: string;\n title: string;\n metric_source: string | null;\n metric_key: string | null;\n start_value: number;\n current_value: number;\n target_value: number;\n unit: string | null;\n confidence: number; // 0..1\n progress: number; // 0..100\n updated_at: string;\n}\n\nexport interface Objective {\n id: string;\n title: string;\n description: string | null;\n quarter: string | null;\n workspace: string | null;\n owner: string | null;\n status: string; // draft | active | done | archived\n progress: number; // 0..100\n created_at: string;\n updated_at: string;\n key_results: KeyResult[];\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp = (n: number, lo: number, hi: number) =>\n Math.max(lo, Math.min(hi, n));\n\nfunction computeKRProgress(kr: KeyResult): number {\n const start = Number(kr.start_value) || 0;\n const cur = Number(kr.current_value) || 0;\n const target = Number(kr.target_value) || 0;\n if (target === start) {\n // Degenerate: target == start. If we're at/above target, call it 100%.\n return cur >= target ? 100 : 0;\n }\n const pct = ((cur - start) / (target - start)) * 100;\n return Math.round(clamp(pct, 0, 100));\n}\n\nfunction computeObjectiveProgress(krs: KeyResult[]): number {\n if (!krs.length) return 0;\n const sum = krs.reduce((acc, k) => acc + (Number(k.progress) || 0), 0);\n return Math.round(sum / krs.length);\n}\n\n// Pulls live integration value for a KR using the unified funnel summary.\n// Mutates kr.current_value + kr.progress in place. Falls through silently on\n// any error so we never break list endpoints because the funnel fetch hiccuped.\nfunction syncKRFromFunnel(\n kr: KeyResult,\n funnelMetrics: Array<{ integration: string; metric: string; value: number }>,\n): void {\n if (!kr.metric_source || !kr.metric_key) {\n kr.progress = computeKRProgress(kr);\n return;\n }\n const hit = funnelMetrics.find(\n (m) => m.integration === kr.metric_source && m.metric === kr.metric_key,\n );\n if (hit && typeof hit.value === \"number\" && Number.isFinite(hit.value)) {\n kr.current_value = Number(hit.value);\n }\n kr.progress = computeKRProgress(kr);\n}\n\nfunction rowToKR(columns: string[], row: any[]): KeyResult {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n objective_id: String(obj.objective_id),\n title: String(obj.title ?? \"\"),\n metric_source: obj.metric_source ?? null,\n metric_key: obj.metric_key ?? null,\n start_value: Number(obj.start_value) || 0,\n current_value: Number(obj.current_value) || 0,\n target_value: Number(obj.target_value) || 0,\n unit: obj.unit ?? null,\n confidence: Number(obj.confidence ?? 0.5),\n progress: Number(obj.progress) || 0,\n updated_at: String(obj.updated_at ?? \"\"),\n };\n}\n\nfunction rowToObjective(columns: string[], row: any[]): Objective {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n quarter: obj.quarter ?? null,\n workspace: obj.workspace ?? null,\n owner: obj.owner ?? null,\n status: String(obj.status ?? \"draft\"),\n progress: Number(obj.progress) || 0,\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n key_results: [],\n };\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listObjectives(\n filters: { workspace?: string; quarter?: string } = {},\n): Promise<Objective[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.workspace) { where.push(\"workspace = ?\"); args.push(filters.workspace); }\n if (filters.quarter) { where.push(\"quarter = ?\"); args.push(filters.quarter); }\n const sql = `SELECT id, title, description, quarter, workspace, owner, status, progress,\n created_at, updated_at\n FROM objectives\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY created_at DESC`;\n const r = await sqlite.execute(sql, args);\n const objectives = (r.rows || []).map(\n (row) => rowToObjective(r.columns, row as any[]),\n );\n if (objectives.length === 0) return [];\n\n // Pull KRs for these objectives in one query.\n const ids = objectives.map((o) => o.id);\n const placeholders = ids.map(() => \"?\").join(\",\");\n const krResult = await sqlite.execute(\n `SELECT id, objective_id, title, metric_source, metric_key, start_value,\n current_value, target_value, unit, confidence, progress, updated_at\n FROM key_results\n WHERE objective_id IN (${placeholders})`,\n ids,\n );\n const krs = (krResult.rows || []).map(\n (row) => rowToKR(krResult.columns, row as any[]),\n );\n\n // Refresh KR current_value from live funnel data for DISPLAY ONLY — no DB writes here.\n // Persisting is the job of syncObjectivesFromFunnel() (cron + POST /api/okrs/sync), so a\n // plain GET stays idempotent (was previously writing key_results + objectives on every read).\n const funnelMetrics = await loadFunnelMetrics();\n for (const kr of krs) syncKRFromFunnel(kr, funnelMetrics);\n\n // Attach KRs + recompute objective progress (in memory).\n const byObj = new Map<string, KeyResult[]>();\n for (const kr of krs) {\n if (!byObj.has(kr.objective_id)) byObj.set(kr.objective_id, []);\n byObj.get(kr.objective_id)!.push(kr);\n }\n for (const obj of objectives) {\n obj.key_results = byObj.get(obj.id) ?? [];\n obj.progress = computeObjectiveProgress(obj.key_results);\n }\n return objectives;\n } catch {\n return [];\n }\n}\n\n// Flatten live funnel data into {integration, metric, value} rows (read-only).\nasync function loadFunnelMetrics(): Promise<Array<{ integration: string; metric: string; value: number }>> {\n try {\n const summary = await getFunnelSummary();\n const rows = summary.stages.flatMap((s) =>\n s.metrics.map((m: any) => ({\n integration: m.integration,\n metric: m.metric,\n value: Number(m.value) || 0,\n })),\n );\n // Revenue wiring: expose CRM/attribution revenue as a funnel metric so a KR with\n // metric_source 'attribution' + metric_key 'revenue' (e.g. the $80K Q2 KR) can actually\n // progress. Won deals call recordConversion → utm_conversions → getAttributionSummary.total_value.\n try {\n const attr = await getAttributionSummary();\n rows.push({ integration: \"attribution\", metric: \"revenue\", value: Number(attr.total_value) || 0 });\n } catch { /* attribution optional */ }\n return rows;\n } catch {\n return [];\n }\n}\n\n// Write path: refresh KR current_value/progress from live funnel data AND persist it, then\n// recompute + persist each objective's progress. Called by the execution cron and by\n// POST /api/okrs/sync — NEVER on a plain GET. Returns how many rows were updated.\nexport async function syncObjectivesFromFunnel(\n filters: { workspace?: string; quarter?: string } = {},\n): Promise<{ objectives: number; key_results: number }> {\n try {\n const objectives = await listObjectives(filters); // read-only; attaches KRs + in-memory funnel sync\n if (objectives.length === 0) return { objectives: 0, key_results: 0 };\n const now = new Date().toISOString();\n let krCount = 0;\n for (const obj of objectives) {\n for (const kr of obj.key_results) {\n try {\n await sqlite.execute(\n \"UPDATE key_results SET current_value = ?, progress = ?, updated_at = ? WHERE id = ?\",\n [kr.current_value, kr.progress, now, kr.id],\n );\n krCount++;\n } catch { /* non-critical */ }\n }\n try {\n await sqlite.execute(\n \"UPDATE objectives SET progress = ?, updated_at = ? WHERE id = ?\",\n [obj.progress, now, obj.id],\n );\n } catch { /* non-critical */ }\n }\n return { objectives: objectives.length, key_results: krCount };\n } catch {\n return { objectives: 0, key_results: 0 };\n }\n}\n\n// ─── Write (Objectives) ──────────────────────────────────────────────────────\n\nexport async function createObjective(\n data: Partial<Objective>,\n): Promise<Objective> {\n const now = new Date().toISOString();\n await sqlite.execute(\n `INSERT INTO objectives\n (title, description, quarter, workspace, owner, status, progress, created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, 0, ?, ?)`,\n [\n data.title ?? \"Untitled Objective\",\n data.description ?? null,\n data.quarter ?? currentQuarter(),\n data.workspace ?? null,\n data.owner ?? null,\n data.status ?? \"draft\",\n now,\n now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return {\n id,\n title: data.title ?? \"Untitled Objective\",\n description: data.description ?? null,\n quarter: data.quarter ?? currentQuarter(),\n workspace: data.workspace ?? null,\n owner: data.owner ?? null,\n status: data.status ?? \"draft\",\n progress: 0,\n created_at: now,\n updated_at: now,\n key_results: [],\n };\n}\n\nexport async function updateObjective(\n id: string,\n data: Partial<Objective>,\n): Promise<Objective | null> {\n try {\n const existing = await sqlite.execute(\n `SELECT id, title, description, quarter, workspace, owner, status, progress,\n created_at, updated_at\n FROM objectives WHERE id = ?`,\n [id],\n );\n if (!existing.rows?.length) return null;\n const cur = rowToObjective(existing.columns, existing.rows[0] as any[]);\n const now = new Date().toISOString();\n const next = {\n title: data.title ?? cur.title,\n description: \"description\" in data ? (data.description ?? null) : cur.description,\n quarter: \"quarter\" in data ? (data.quarter ?? null) : cur.quarter,\n workspace: \"workspace\" in data ? (data.workspace ?? null) : cur.workspace,\n owner: \"owner\" in data ? (data.owner ?? null) : cur.owner,\n status: data.status ?? cur.status,\n };\n await sqlite.execute(\n `UPDATE objectives\n SET title = ?, description = ?, quarter = ?, workspace = ?, owner = ?,\n status = ?, updated_at = ?\n WHERE id = ?`,\n [next.title, next.description, next.quarter, next.workspace, next.owner, next.status, now, id],\n );\n const list = await listObjectives({});\n return list.find((o) => o.id === id) ?? null;\n } catch {\n return null;\n }\n}\n\nexport async function deleteObjective(id: string): Promise<boolean> {\n try {\n const check = await sqlite.execute(\"SELECT id FROM objectives WHERE id = ?\", [id]);\n if (!check.rows?.length) return false;\n // Manual cascade in case FK enforcement is off.\n await sqlite.execute(\"DELETE FROM key_results WHERE objective_id = ?\", [id]);\n await sqlite.execute(\"DELETE FROM objectives WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── Write (Key Results) ─────────────────────────────────────────────────────\n\nexport async function createKeyResult(\n objective_id: string,\n data: Partial<KeyResult>,\n): Promise<KeyResult> {\n const now = new Date().toISOString();\n const start_value = Number(data.start_value) || 0;\n const current_value = Number(data.current_value ?? data.start_value ?? 0);\n const target_value = Number(data.target_value) || 0;\n // QA fix: auto-normalize confidence — chat tools + UI sliders may send 1-10 instead of\n // the schema's 0-1. Detect any value > 1 and divide by 10.\n const rawConf = Number(data.confidence ?? 0.5);\n const confidence = clamp(rawConf > 1 ? rawConf / 10 : rawConf, 0, 1);\n\n const initial: KeyResult = {\n id: \"0\", // placeholder, replaced after insert\n objective_id,\n title: data.title ?? \"Untitled KR\",\n metric_source: data.metric_source ?? null,\n metric_key: data.metric_key ?? null,\n start_value,\n current_value,\n target_value,\n unit: data.unit ?? null,\n confidence,\n progress: 0,\n updated_at: now,\n };\n initial.progress = computeKRProgress(initial);\n\n await sqlite.execute(\n `INSERT INTO key_results\n (objective_id, title, metric_source, metric_key, start_value, current_value,\n target_value, unit, confidence, progress, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n initial.objective_id, initial.title, initial.metric_source,\n initial.metric_key, initial.start_value, initial.current_value,\n initial.target_value, initial.unit, initial.confidence, initial.progress, now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n initial.id = String((rowIdResult.rows[0] as any[])[0]);\n return initial;\n}\n\nexport async function updateKeyResult(\n id: string,\n data: Partial<KeyResult>,\n): Promise<KeyResult | null> {\n try {\n const existing = await sqlite.execute(\n `SELECT id, objective_id, title, metric_source, metric_key, start_value,\n current_value, target_value, unit, confidence, progress, updated_at\n FROM key_results WHERE id = ?`,\n [id],\n );\n if (!existing.rows?.length) return null;\n const cur = rowToKR(existing.columns, existing.rows[0] as any[]);\n const next: KeyResult = {\n ...cur,\n title: data.title ?? cur.title,\n metric_source: \"metric_source\" in data ? (data.metric_source ?? null) : cur.metric_source,\n metric_key: \"metric_key\" in data ? (data.metric_key ?? null) : cur.metric_key,\n start_value: data.start_value != null ? Number(data.start_value) : cur.start_value,\n current_value: data.current_value != null ? Number(data.current_value) : cur.current_value,\n target_value: data.target_value != null ? Number(data.target_value) : cur.target_value,\n unit: \"unit\" in data ? (data.unit ?? null) : cur.unit,\n confidence: data.confidence != null\n ? clamp((() => { const c = Number(data.confidence); return c > 1 ? c / 10 : c; })(), 0, 1)\n : cur.confidence,\n updated_at: new Date().toISOString(),\n };\n next.progress = computeKRProgress(next);\n await sqlite.execute(\n `UPDATE key_results\n SET title = ?, metric_source = ?, metric_key = ?, start_value = ?,\n current_value = ?, target_value = ?, unit = ?, confidence = ?,\n progress = ?, updated_at = ?\n WHERE id = ?`,\n [\n next.title, next.metric_source, next.metric_key, next.start_value,\n next.current_value, next.target_value, next.unit, next.confidence,\n next.progress, next.updated_at, id,\n ],\n );\n return next;\n } catch {\n return null;\n }\n}\n\n// ─── Dashboard ───────────────────────────────────────────────────────────────\n\nexport async function getOKRDashboard(): Promise<{\n total: number;\n on_track: number;\n at_risk: number;\n avg_progress: number;\n quarter: string | null;\n}> {\n const fallback = { total: 0, on_track: 0, at_risk: 0, avg_progress: 0, quarter: null as string | null };\n try {\n const objectives = await listObjectives({});\n if (!objectives.length) return fallback;\n\n let on_track = 0;\n let at_risk = 0;\n let progressSum = 0;\n let progressCount = 0;\n for (const o of objectives) {\n // Aggregate KR confidence per objective so we can compare progress vs\n // expected progress at that confidence level.\n const avgConfidence = o.key_results.length\n ? o.key_results.reduce((a, k) => a + Number(k.confidence || 0), 0) /\n o.key_results.length\n : 0.5;\n const expected = avgConfidence * 100;\n if (o.progress >= expected) on_track += 1;\n if (o.progress < expected - 15) at_risk += 1;\n progressSum += o.progress;\n progressCount += 1;\n }\n const avg_progress = progressCount > 0\n ? Math.round(progressSum / progressCount)\n : 0;\n\n // Pick the most common quarter as the current quarter label.\n const counts = new Map<string, number>();\n for (const o of objectives) {\n if (!o.quarter) continue;\n counts.set(o.quarter, (counts.get(o.quarter) || 0) + 1);\n }\n let quarter: string | null = null;\n let max = 0;\n for (const [q, n] of counts) {\n if (n > max) { max = n; quarter = q; }\n }\n // QA fix: when there are no objectives, fall back to the CURRENT quarter computed\n // from today's date rather than returning null (which the UI was rendering as \"—\").\n if (!quarter) {\n const now = new Date();\n const q = Math.floor(now.getUTCMonth() / 3) + 1;\n quarter = `Q${q}-${now.getUTCFullYear()}`;\n }\n\n return { total: objectives.length, on_track, at_risk, avg_progress, quarter };\n } catch {\n return { ...fallback, quarter: (() => {\n const now = new Date();\n const q = Math.floor(now.getUTCMonth() / 3) + 1;\n return `Q${q}-${now.getUTCFullYear()}`;\n })() };\n }\n}\n",[],[365,746],"29b479bc-5d69-11f1-98bb-1607ee4eb77e","nsm.ts","964e9fd2-589e-11f1-bbad-ee650bb23af1",620,["D",1780284167768],"// North Star Metric — Phase 1 Framework Layer.\n// Singleton NSM row (one per workspace at a time) + rolling history for trend\n// + velocity. Auto-pulls live value from the funnel summary when the NSM is\n// pinned to a connected integration metric (e.g. ga4:conversions).\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp = (n: number, lo: number, hi: number) =>\n Math.max(lo, Math.min(hi, n));\n\nconst daysBetween = (laterISO: string, earlierISO: string) =>\n (new Date(laterISO).getTime() - new Date(earlierISO).getTime()) / 86400000;\n\nexport interface NSMRow {\n id: string;\n metric_name: string;\n metric_source: string | null;\n metric_key: string | null;\n current_value: number;\n target_value: number;\n target_date: string | null;\n unit: string | null;\n description: string | null;\n set_at: string;\n set_by: string | null;\n}\n\n// Resolve a live metric value from the unified funnel summary when the NSM is\n// pinned to a known integration metric. Returns null when there's no match\n// (e.g. for a manually-tracked north star like \"revenue\" with no integration).\nasync function resolveLiveValue(\n metric_source: string | null,\n metric_key: string | null,\n): Promise<number | null> {\n if (!metric_source || !metric_key) return null;\n try {\n const summary = await getFunnelSummary();\n const metrics = summary.stages.flatMap((s) => s.metrics);\n const hit = metrics.find(\n (m: any) => m.integration === metric_source && m.metric === metric_key,\n );\n if (hit && typeof (hit as any).value === \"number\") {\n return Number((hit as any).value);\n }\n return null;\n } catch {\n return null;\n }\n}\n\n// ─── CRUD ────────────────────────────────────────────────────────────────────\n\nexport async function getNSM(): Promise<\n { nsm: NSMRow | null; history: Array<{ value: number; recorded_at: string }> }\n> {\n try {\n const r = await sqlite.execute(\n `SELECT id, metric_name, metric_source, metric_key, current_value, target_value,\n target_date, unit, description, set_at, set_by\n FROM north_star\n ORDER BY set_at DESC\n LIMIT 1`,\n );\n if (!r.rows?.length) return { nsm: null, history: [] };\n const obj = rowToObj(r.columns, r.rows[0] as any[]);\n\n // Sync live value if pinned to a connected integration.\n const live = await resolveLiveValue(obj.metric_source, obj.metric_key);\n if (live !== null && Number.isFinite(live) && live !== Number(obj.current_value)) {\n try {\n await sqlite.execute(\n \"UPDATE north_star SET current_value = ? WHERE id = ?\",\n [live, obj.id],\n );\n obj.current_value = live;\n // Append to history so the cron/dashboard sees the refreshed point.\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, datetime('now'))\",\n [obj.id, live],\n );\n } catch { /* non-critical */ }\n }\n\n const nsm: NSMRow = {\n id: String(obj.id),\n metric_name: String(obj.metric_name ?? \"\"),\n metric_source: obj.metric_source ?? null,\n metric_key: obj.metric_key ?? null,\n current_value: Number(obj.current_value) || 0,\n target_value: Number(obj.target_value) || 0,\n target_date: obj.target_date ?? null,\n unit: obj.unit ?? null,\n description: obj.description ?? null,\n set_at: String(obj.set_at ?? new Date().toISOString()),\n set_by: obj.set_by ?? null,\n };\n\n const hist = await sqlite.execute(\n `SELECT value, recorded_at\n FROM nsm_history\n WHERE nsm_id = ?\n AND recorded_at >= datetime('now','-90 days')\n ORDER BY recorded_at ASC`,\n [nsm.id],\n );\n const history = (hist.rows || []).map((row: any) => ({\n value: Number(row[0]) || 0,\n recorded_at: String(row[1]),\n }));\n\n return { nsm, history };\n } catch {\n return { nsm: null, history: [] };\n }\n}\n\nexport async function setNSM(data: {\n metric_name: string;\n metric_source?: string | null;\n metric_key?: string | null;\n target_value: number;\n target_date?: string | null;\n unit?: string | null;\n description?: string | null;\n set_by?: string | null;\n}): Promise<NSMRow> {\n const now = new Date().toISOString();\n\n // Singleton semantics: clear out previous NSM(s) + their history.\n try {\n const prior = await sqlite.execute(\"SELECT id FROM north_star\");\n for (const row of (prior.rows || []) as any[]) {\n const priorId = String(row[0]);\n await sqlite.execute(\"DELETE FROM nsm_history WHERE nsm_id = ?\", [priorId]);\n }\n await sqlite.execute(\"DELETE FROM north_star\");\n } catch { /* non-critical */ }\n\n // Seed current_value from the live integration value when pinned, else 0.\n const live = await resolveLiveValue(\n data.metric_source ?? null,\n data.metric_key ?? null,\n );\n const current_value = live !== null && Number.isFinite(live) ? live : 0;\n const target_date = data.target_date ?? now.slice(0, 10);\n\n await sqlite.execute(\n `INSERT INTO north_star\n (metric_name, metric_source, metric_key, current_value, target_value,\n target_date, unit, description, set_at, set_by)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.metric_name,\n data.metric_source ?? \"\", // NOT NULL column — default to empty string for manual NSMs\n data.metric_key ?? null,\n current_value,\n Number(data.target_value) || 0,\n target_date,\n data.unit ?? null,\n data.description ?? null,\n now,\n data.set_by ?? null,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n\n // Seed nsm_history only when we already have a non-zero starting value (e.g. pulled live\n // from the connected integration). User QA flagged a 0-value blip when this fired before\n // the first metric sync — that blip dragged velocity calculations to -∞.\n if (current_value > 0) {\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, ?)\",\n [id, current_value, now],\n );\n }\n\n return {\n id,\n metric_name: data.metric_name,\n metric_source: data.metric_source ?? \"\",\n metric_key: data.metric_key ?? null,\n current_value,\n target_value: Number(data.target_value) || 0,\n target_date,\n unit: data.unit ?? null,\n description: data.description ?? null,\n set_at: now,\n set_by: data.set_by ?? null,\n };\n}\n\nexport async function recordNSMValue(value: number): Promise<void> {\n try {\n const r = await sqlite.execute(\n \"SELECT id FROM north_star ORDER BY set_at DESC LIMIT 1\",\n );\n if (!r.rows?.length) return;\n const nsmId = String((r.rows[0] as any[])[0]);\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, datetime('now'))\",\n [nsmId, Number(value) || 0],\n );\n await sqlite.execute(\n \"UPDATE north_star SET current_value = ? WHERE id = ?\",\n [Number(value) || 0, nsmId],\n );\n } catch { /* non-critical */ }\n}\n\n// ─── Progress / Velocity ─────────────────────────────────────────────────────\n\nexport interface NSMProgress {\n current: number;\n target: number;\n target_date: string | null;\n pct_complete: number; // 0..100\n days_remaining: number | null; // until target_date\n on_track: boolean;\n velocity_per_day: number | null;\n days_to_target: number | null; // null when velocity ≤ 0 or no history\n}\n\nexport async function getNSMProgress(): Promise<NSMProgress> {\n const fallback: NSMProgress = {\n current: 0,\n target: 0,\n target_date: null,\n pct_complete: 0,\n days_remaining: null,\n on_track: false,\n velocity_per_day: null,\n days_to_target: null,\n };\n\n try {\n const { nsm, history } = await getNSM();\n if (!nsm) return fallback;\n\n const current = Number(nsm.current_value) || 0;\n const target = Number(nsm.target_value) || 0;\n\n // Velocity = (current - value_30d_ago) / 30. Use the oldest sample within\n // the trailing 30d window when an exact 30d-old point isn't present.\n let value30dAgo: number | null = null;\n if (history.length > 0) {\n const thirtyAgo = Date.now() - 30 * 86400000;\n // Earliest sample >= 30d ago, or fall back to the first sample we have.\n const inWindow = history.filter(\n (h) => new Date(h.recorded_at).getTime() <= thirtyAgo,\n );\n if (inWindow.length > 0) {\n value30dAgo = inWindow[inWindow.length - 1].value;\n } else {\n value30dAgo = history[0].value;\n }\n }\n\n const velocity_per_day =\n value30dAgo !== null ? (current - value30dAgo) / 30 : null;\n\n const days_to_target =\n velocity_per_day !== null && velocity_per_day > 0\n ? (target - current) / velocity_per_day\n : null;\n\n const days_remaining =\n nsm.target_date\n ? Math.max(0, daysBetween(nsm.target_date, new Date().toISOString()))\n : null;\n\n const on_track =\n days_to_target !== null &&\n days_remaining !== null &&\n days_to_target <= days_remaining;\n\n const pct_complete = target > 0 ? clamp((current / target) * 100, 0, 100) : 0;\n\n return {\n current,\n target,\n target_date: nsm.target_date,\n pct_complete,\n days_remaining,\n on_track,\n velocity_per_day,\n days_to_target,\n };\n } catch {\n return fallback;\n }\n}\n",[],[365,754],"6973f69c-5989-11f1-aff7-ee650bb23af1","method_config.ts","6973f750-5989-11f1-aff7-ee650bb23af1",366,["D",1779858213970],"// Configurable Metrics + Growth Methodology.\n// Backs the Settings → \"Growth Methodology\" panel. Lets users:\n// 1) Pick a prioritization method (ICE | RICE | WSJF) — drives suggestICE + experiment scoring.\n// 2) Override funnel-stage metrics — e.g. swap GSC clicks → impressions on TOFU.\n// 3) Define custom KPIs (name + formula + unit) that surface on Analytics.\n//\n// Storage uses the existing `app_config` (key/value JSON) table — no new schema.\nimport { sqlite } from \"../db.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type PrioritizationMethod = \"ICE\" | \"RICE\" | \"WSJF\";\n\nexport interface MetricOverride {\n integration: string; // e.g. \"gsc\", \"ga4\"\n metric: string; // e.g. \"impressions\", \"ctr\"\n weight: number; // 0..2 — multiplier applied during health computation\n}\n\nexport interface CustomKPI {\n name: string; // user-defined label\n formula: string; // free-form expression (eval'd in Analytics layer; stored as text)\n unit: string; // \"count\", \"%\", \"$\", \"s\", etc.\n}\n\nexport interface FunnelMetricOverrides {\n TOFU: MetricOverride[];\n MOFU: MetricOverride[];\n BOFU: MetricOverride[];\n}\n\nexport interface MethodConfig {\n prioritization_method: PrioritizationMethod;\n funnel_metric_overrides: FunnelMetricOverrides;\n custom_kpis: CustomKPI[];\n}\n\n// ─── Static method metadata ──────────────────────────────────────────────────\n// `compute` returns the prioritization score; identical contract regardless of method\n// so callers can do `METHODS[m].compute(...args)` without branching.\n\nexport const METHODS = {\n ICE: {\n name: \"ICE\",\n formula: \"(Impact + Confidence + Ease) / 3\",\n range: \"0-10 each\",\n fields: [\"impact\", \"confidence\", \"ease\"] as string[],\n defaults: { impact: 5, confidence: 5, ease: 5 },\n compute: (i: number, c: number, e: number): number =>\n Math.round(((i + c + e) / 3) * 100) / 100,\n why:\n \"Fast triage. Best when you have many ideas and need to ship quickly. \" +\n \"Used by Sean Ellis at LogMeIn/Dropbox.\",\n },\n RICE: {\n name: \"RICE\",\n formula: \"(Reach × Impact × Confidence) / Effort\",\n range:\n \"Reach: # users/period, Impact: 0.25-3, Confidence: 0-100%, Effort: person-weeks\",\n fields: [\"reach\", \"impact\", \"confidence\", \"effort\"] as string[],\n defaults: { reach: 1000, impact: 1, confidence: 0.8, effort: 2 },\n compute: (r: number, i: number, c: number, e: number): number =>\n Math.round(((r * i * c) / Math.max(e, 0.1)) * 100) / 100,\n why:\n \"Better when reach varies a lot across experiments. Used by Intercom.\",\n },\n WSJF: {\n name: \"WSJF (Weighted Shortest Job First)\",\n formula: \"(Business Value + Time Criticality + Risk Reduction) / Job Size\",\n range: \"1-10 each (modified Fibonacci)\",\n fields: [\"business_value\", \"time_criticality\", \"risk_reduction\", \"job_size\"] as string[],\n defaults: { business_value: 5, time_criticality: 5, risk_reduction: 5, job_size: 3 },\n compute: (bv: number, tc: number, rr: number, js: number): number =>\n Math.round(((bv + tc + rr) / Math.max(js, 1)) * 100) / 100,\n why:\n \"SAFe / agile teams. Best when ladder of work item sizes matters.\",\n },\n} as const;\n\nexport type MethodKey = keyof typeof METHODS;\n\n// ─── Persistence helpers ─────────────────────────────────────────────────────\n\nconst KEY_METHOD = \"prioritization_method\";\nconst KEY_OVERRIDES = \"funnel_metric_overrides\";\nconst KEY_KPIS = \"custom_kpis\";\n\nasync function getConfigValue(key: string): Promise<string | null> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [key]);\n if (!r.rows?.length) return null;\n const v = (r.rows[0] as any[])[0];\n return v == null ? null : String(v);\n } catch {\n return null;\n }\n}\n\nasync function setConfigValue(key: string, value: string): Promise<void> {\n const now = new Date().toISOString();\n try {\n await sqlite.execute(\n \"INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, ?) \" +\n \"ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = excluded.updated_at\",\n [key, value, now],\n );\n } catch {\n // Fallback for sqlite builds without UPSERT — manual update-or-insert.\n try {\n const existing = await sqlite.execute(\"SELECT key FROM app_config WHERE key = ?\", [key]);\n if (existing.rows?.length) {\n await sqlite.execute(\"UPDATE app_config SET value = ?, updated_at = ? WHERE key = ?\", [value, now, key]);\n } else {\n await sqlite.execute(\"INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, ?)\", [key, value, now]);\n }\n } catch { /* swallow — non-critical */ }\n }\n}\n\nfunction normalizeMethod(raw: any): PrioritizationMethod {\n const s = String(raw ?? \"\").toUpperCase();\n if (s === \"RICE\" || s === \"WSJF\" || s === \"ICE\") return s as PrioritizationMethod;\n return \"ICE\";\n}\n\nfunction normalizeOverrides(raw: any): FunnelMetricOverrides {\n const empty: FunnelMetricOverrides = { TOFU: [], MOFU: [], BOFU: [] };\n if (!raw || typeof raw !== \"object\") return empty;\n const out: FunnelMetricOverrides = { TOFU: [], MOFU: [], BOFU: [] };\n for (const stage of [\"TOFU\", \"MOFU\", \"BOFU\"] as const) {\n const arr = Array.isArray(raw[stage]) ? raw[stage] : [];\n out[stage] = arr\n .filter((r: any) => r && typeof r === \"object\" && r.integration && r.metric)\n .map((r: any) => ({\n integration: String(r.integration),\n metric: String(r.metric),\n weight: Number.isFinite(Number(r.weight)) ? Number(r.weight) : 1,\n }));\n }\n return out;\n}\n\nfunction normalizeKpis(raw: any): CustomKPI[] {\n if (!Array.isArray(raw)) return [];\n return raw\n .filter((k: any) => k && typeof k === \"object\" && k.name && k.formula)\n .map((k: any) => ({\n name: String(k.name).slice(0, 80),\n formula: String(k.formula).slice(0, 500),\n unit: String(k.unit ?? \"count\").slice(0, 20),\n }));\n}\n\n// ─── Public API ──────────────────────────────────────────────────────────────\n\nexport async function getMethodConfig(): Promise<MethodConfig> {\n const [m, o, k] = await Promise.all([\n getConfigValue(KEY_METHOD),\n getConfigValue(KEY_OVERRIDES),\n getConfigValue(KEY_KPIS),\n ]);\n let overridesRaw: any = null;\n let kpisRaw: any = null;\n try { if (o) overridesRaw = JSON.parse(o); } catch { /* malformed */ }\n try { if (k) kpisRaw = JSON.parse(k); } catch { /* malformed */ }\n return {\n prioritization_method: normalizeMethod(m),\n funnel_metric_overrides: normalizeOverrides(overridesRaw),\n custom_kpis: normalizeKpis(kpisRaw),\n };\n}\n\nexport async function saveMethodConfig(patch: Partial<MethodConfig>): Promise<MethodConfig> {\n const current = await getMethodConfig();\n const next: MethodConfig = {\n prioritization_method: patch.prioritization_method !== undefined\n ? normalizeMethod(patch.prioritization_method)\n : current.prioritization_method,\n funnel_metric_overrides: patch.funnel_metric_overrides !== undefined\n ? normalizeOverrides(patch.funnel_metric_overrides)\n : current.funnel_metric_overrides,\n custom_kpis: patch.custom_kpis !== undefined\n ? normalizeKpis(patch.custom_kpis)\n : current.custom_kpis,\n };\n await Promise.all([\n setConfigValue(KEY_METHOD, next.prioritization_method),\n setConfigValue(KEY_OVERRIDES, JSON.stringify(next.funnel_metric_overrides)),\n setConfigValue(KEY_KPIS, JSON.stringify(next.custom_kpis)),\n ]);\n return next;\n}\n\n// Helper used by api/experiments.ts to fetch JUST the method (cheaper, no JSON parse).\nexport async function getPrioritizationMethod(): Promise<PrioritizationMethod> {\n return normalizeMethod(await getConfigValue(KEY_METHOD));\n}\n\n// Static metadata payload for GET /api/method-config/methods — surfaces formula + why-to-use\n// in the Settings UI without round-trips. Strips the `compute` function (non-serializable).\nexport function getMethodsMetadata(): Array<{\n key: MethodKey;\n name: string;\n formula: string;\n range: string;\n fields: string[];\n defaults: Record<string, number>;\n why: string;\n}> {\n return (Object.keys(METHODS) as MethodKey[]).map((key) => {\n const m = METHODS[key];\n return {\n key,\n name: m.name,\n formula: m.formula,\n range: m.range,\n fields: [...m.fields],\n defaults: { ...m.defaults },\n why: m.why,\n };\n });\n}\n",[],[365,762],"d48917c4-5d61-11f1-9b76-1607ee4eb77e","memory.ts","d489186e-5d61-11f1-9b76-1607ee4eb77e",578,["D",1780281018405],"// Part 15.3 — Memory: persistent, evolving business memory.\n// Company Profile = static identity; Memory = evolving learned context the AI accumulates and\n// reuses across conversations. CRUD + retrieval + a compact injection string + a best-effort\n// auto-capture that distills durable facts from a finished conversation.\nimport { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\nexport interface MemoryItem {\n id: number;\n type: string; // company | preference | fact | decision | learning\n content: string;\n source: string; // chat | agent | experiment | manual\n confidence: number;\n pinned: boolean;\n created_at: string;\n updated_at: string;\n}\n\nconst VALID_TYPES = new Set([\"company\", \"preference\", \"fact\", \"decision\", \"learning\"]);\n\nfunction rowTo(cols: string[], row: any[]): MemoryItem {\n const o: any = {};\n for (let i = 0; i < cols.length; i++) o[cols[i]] = row[i];\n return { ...o, pinned: !!o.pinned, confidence: Number(o.confidence) };\n}\n\nexport async function listMemory(): Promise<MemoryItem[]> {\n try {\n const r = await sqlite.execute(\n \"SELECT id, type, content, source, confidence, pinned, created_at, updated_at FROM memory ORDER BY pinned DESC, updated_at DESC LIMIT 200\"\n );\n return (r.rows || []).map((row) => rowTo(r.columns, row as any[]));\n } catch { return []; }\n}\n\nfunction normContent(s: string): string {\n return s.trim().toLowerCase().replace(/\\s+/g, \" \");\n}\n\nexport async function addMemory(data: { content: string; type?: string; source?: string; confidence?: number; pinned?: boolean }): Promise<MemoryItem | null> {\n const content = String(data.content || \"\").trim();\n if (!content) return null;\n const type = data.type && VALID_TYPES.has(data.type) ? data.type : \"fact\";\n const source = (data.source || \"manual\").toString();\n const confidence = typeof data.confidence === \"number\" ? Math.max(0, Math.min(1, data.confidence)) : 0.7;\n const pinned = data.pinned ? 1 : 0;\n\n // Dedupe: if an existing memory has the same normalized content, bump it instead of duplicating.\n try {\n const existing = await sqlite.execute(\"SELECT id, content FROM memory\");\n for (const row of (existing.rows || []) as any[]) {\n if (normContent(String(row[1])) === normContent(content)) {\n await sqlite.execute(\"UPDATE memory SET updated_at = datetime('now'), confidence = ? WHERE id = ?\", [Math.max(confidence, 0.7), row[0]]);\n return await getMemory(Number(row[0]));\n }\n }\n } catch { /* table optional */ }\n\n await sqlite.execute(\n \"INSERT INTO memory (type, content, source, confidence, pinned) VALUES (?, ?, ?, ?, ?)\",\n [type, content, source, confidence, pinned]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n return await getMemory(Number((idRow.rows[0] as any[])[0]));\n}\n\nexport async function getMemory(id: number): Promise<MemoryItem | null> {\n const r = await sqlite.execute(\n \"SELECT id, type, content, source, confidence, pinned, created_at, updated_at FROM memory WHERE id = ?\", [id]\n );\n if (!r.rows?.length) return null;\n return rowTo(r.columns, r.rows[0] as any[]);\n}\n\nexport async function updateMemory(id: number, data: { content?: string; type?: string; pinned?: boolean; confidence?: number }): Promise<MemoryItem | null> {\n const cur = await getMemory(id);\n if (!cur) return null;\n const content = typeof data.content === \"string\" ? data.content.trim() : cur.content;\n const type = data.type && VALID_TYPES.has(data.type) ? data.type : cur.type;\n const pinned = typeof data.pinned === \"boolean\" ? (data.pinned ? 1 : 0) : (cur.pinned ? 1 : 0);\n const confidence = typeof data.confidence === \"number\" ? Math.max(0, Math.min(1, data.confidence)) : cur.confidence;\n await sqlite.execute(\n \"UPDATE memory SET content = ?, type = ?, pinned = ?, confidence = ?, updated_at = datetime('now') WHERE id = ?\",\n [content, type, pinned, confidence, id]\n );\n return await getMemory(id);\n}\n\nexport async function deleteMemory(id: number): Promise<boolean> {\n try { await sqlite.execute(\"DELETE FROM memory WHERE id = ?\", [id]); return true; } catch { return false; }\n}\n\n// Compact injection: pinned first, then most-recent, capped (~2KB per Part 15.4 budget).\nexport async function getMemoryContextString(maxChars = 2000): Promise<string> {\n const items = await listMemory();\n if (!items.length) return \"\";\n let out = \"## Memory (what we've learned about this business)\\n\";\n let remaining = maxChars - out.length;\n for (const m of items) {\n const line = `- [${m.type}] ${m.content}\\n`;\n if (line.length > remaining) break;\n out += line;\n remaining -= line.length;\n }\n return out.trim();\n}\n\n// Best-effort auto-capture: distill durable facts from a finished conversation. Fire-and-forget;\n// never block the chat response. Only runs when an LLM key exists and there's enough signal.\nexport async function captureMemoryFromConversation(messages: Array<{ role: string; content: string }>): Promise<void> {\n try {\n const convo = messages.filter((m) => m.role === \"user\" || m.role === \"assistant\")\n .map((m) => `${m.role}: ${String(m.content || \"\").slice(0, 1200)}`).join(\"\\n\");\n if (convo.length < 200) return; // too little to learn from\n const cfg = await getLLMConfig();\n if (!(cfg as any).api_key) return;\n\n const prompt = `From the conversation below, extract 0-3 DURABLE facts worth remembering about the\nuser's BUSINESS (not transient task details). Examples of durable: target market/ICP, tone preference,\na stated goal/target, a strategic decision, a learned constraint. Ignore one-off questions and pleasantries.\nReturn ONLY a JSON array of objects: [{\"type\":\"company|preference|fact|decision|learning\",\"content\":\"<concise fact>\"}].\nReturn [] if nothing is durable.\n\nCONVERSATION:\n${convo.slice(0, 6000)}`;\n\n const resp = await callLLM(cfg, [{ role: \"user\", content: prompt }], { maxTokens: 400 });\n const text = (resp.content || \"\").trim();\n const match = text.match(/\\[[\\s\\S]*\\]/);\n if (!match) return;\n let arr: any[] = [];\n try { arr = JSON.parse(match[0]); } catch { return; }\n for (const item of arr.slice(0, 3)) {\n if (item && typeof item.content === \"string\" && item.content.trim()) {\n await addMemory({ content: item.content, type: item.type, source: \"chat\", confidence: 0.6 });\n }\n }\n } catch { /* non-fatal */ }\n}\n",[],[365,770],"bdb7148e-5a4f-11f1-9df7-ee650bb23af1","math_knowledge_seed.ts","3acb1cf0-5991-11f1-8131-ee650bb23af1",404,["D",1779943395684],"// Math/framework knowledge seeds for Nexus.\n// Loaded into knowledge_files on init. LLM prompts inject\n// rows whose `targets` match the active feature. Empty = global.\n\nexport interface MathKnowledgeSeed {\n name: string; // human label\n kind: 'formula' | 'framework' | 'benchmark';\n targets: string[]; // which AI features consume this\n content: string; // 400-800 words of dense, AI-readable markdown\n}\n\nexport const MATH_KNOWLEDGE_SEEDS: MathKnowledgeSeed[] = [\n {\n name: 'Customer Acquisition Cost (CAC)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest'],\n content: `Source: SaaS unit economics (David Skok, Bessemer State of the Cloud).\n\n### Formula\n- CAC = est_marketing_spend / new_customers_acquired\n- Window: trailing 30 or 90 days. Spend window MUST match customer window.\n- Fully-loaded CAC = (marketing + sales_payroll + tooling + agency) / new_customers. Nexus uses paid-channel spend only — state this caveat when comparing to public benchmarks.\n- Blended CAC = total_spend / total_new_customers. Channel CAC = channel_spend / channel_attributed_customers.\n\n### Thresholds\n- Payback period = CAC / (ARPU × gross_margin). Healthy SaaS: < 12 months. World-class: < 6 months.\n- CAC must be < LTV / 3. If CAC ≥ LTV/3 → UNSUSTAINABLE.\n- B2B SaaS medians: $700 SMB, $7k mid-market, $25k+ enterprise (OpenView 2024).\n- Ecommerce: blended CAC ≤ 30% of first-order AOV for repeat categories.\n\n### When/Then\n- CAC trend rises > 20% MoM AND new_customers flat → channel saturation / creative fatigue. Action: rotate creative, test new channel.\n- CAC rises and LTV rises proportionally → healthy if LTV/CAC ≥ 3 holds.\n- CAC > LTV/3 → flag UNSUSTAINABLE in digest. Recommend pausing worst-MER channel.\n- Channel CAC < blended × 0.7 → scale that channel.\n- Channel CAC > blended × 1.5 → cut or restructure.\n- Rising CAC + falling MER + flat conversion = paid inefficiency (creative/audience).\n- Rising CAC + falling conversion = funnel quality issue (landing page / ICP).\n\n### Anti-pattern\n- Comparing CAC across different window lengths — always normalize.\n- Excluding organic customers from denominator while including all spend → inflated CAC.\n- Ignoring sales-cycle lag: customers this month caused by spend ~lag days ago. Lag-correct when sales_cycle > window.\n\n### Interpretation for chat\n- \"Is our CAC OK?\" → answer with current CAC, LTV/CAC ratio, payback period. Verdict only if all three computable.\n- If CAC missing, do not fabricate. State which input is missing and which integration supplies it (Stripe/HubSpot for customers, Meta/Google/LinkedIn for spend).\n- Rising trend: 3-week rolling vs prior 3-week. > 15% = \"rising,\" > 30% = \"spiking.\" Cross-check seasonality / campaign launches first.\n\n### Decision rules (Nexus)\n- payback > 18mo → digest \"Cash-efficiency risk.\"\n- LTV/CAC < 3 → recommendations: \"Tighten ICP\" or \"Cut worst channel.\"\n- CAC trend down + LTV trend up → digest \"Compounding efficiency.\"\n- Channel reallocation only if delta vs blended > 30% AND channel spend > $1k/week (noise filter).\n- One paid channel → blended only; recommend channel diversification.\n- CAC from < 14d data → label \"preliminary\"; require ≥ 30d before automated alerts.`,\n },\n\n {\n name: 'Lifetime Value (LTV)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest', 'cohort_ltv'],\n content: `Source: David Skok (For Entrepreneurs), Bessemer cohort LTV.\n\n### Formula\n- Simple LTV = avg_deal_size × retention_months\n- Margin-adjusted LTV = ARPU × gross_margin × (1 / monthly_churn)\n- Cohort survival LTV = ARPU × Σ retention_rate_t for t=1..N. Sum the observed cohort survival curve; do NOT extrapolate beyond observed data unless fitting a curve (geometric / Kaplan-Meier).\n- Discounted LTV = Σ(ARPU_t × retention_t / (1+r)^t).\n\n### Thresholds\n- LTV/CAC ≥ 3 = healthy. ≥ 5 = world-class (Bessemer top quartile). < 1 = burning cash per customer.\n- NRR benchmark: ≥ 100% (best-in-class ≥ 120%).\n- GRR benchmark: ≥ 90% SMB, ≥ 95% mid+.\n- Monthly logo churn: < 1% excellent, 1-3% acceptable SMB, > 5% critical.\n\n### When/Then\n- LTV flat 3 months AND CAC rising → CAC efficiency degrading. Focus on retention/expansion before more acquisition.\n- LTV rises via expansion (NRR > 100%) → double acquisition spend; payback shortens.\n- LTV rises but only top-decile customers expand → median LTV may be falling. Segment-level analysis required.\n- Cohort survival curves diverge by channel → channel quality differs. Allocate to high-LTV channels even at higher CAC.\n- Monthly churn > 5% → cap LTV reporting; LTV math unstable (sensitive to small churn shifts).\n\n### Cohort survival LTV (api/aio.ts)\n- For each cohort C_m, retention_t = active(m+t) / cohort_size.\n- LTV(C_m, N) = ARPU(C_m) × Σ retention_t for t=0..N. Default N=24mo.\n- Report curve shape: \"flattening after month 6\" = sticky, \"linear decay\" = transactional.\n- Cohorts < 6 months: label \"preliminary\"; use blended curve as prior.\n\n### Anti-pattern\n- 1/churn with < 12 months data — extrapolation explodes (1/1% = 100 months projected).\n- Mixing gross/net revenue — pick one.\n- Ignoring gross margin. $100 ARPU at 30% margin → effective $30 LTV/mo.\n- Single blended LTV across SMB+Enterprise — segment when divergence > 2x.\n\n### Interpretation for chat\n- \"What's our LTV?\" → return ARPU, gross margin, churn, resulting LTV. State which formula (simple / cohort / margin-adjusted).\n- LTV is a forecast, not a fact. Communicate uncertainty: cohort age + sample size.\n- Refuse single LTV number when segments diverge > 2x — break it down.\n\n### Decision rules\n- LTV/CAC < 3 + churn > 5% → recommendations: \"Retention sprint\" over acquisition.\n- LTV trend up 10%+ / 90d AND CAC flat → digest \"Compounding.\"\n- Cohort curve flattens by month 3 → PMF signal; recommend acquisition scale-up.`,\n },\n\n {\n name: 'Marketing Efficiency Ratio (MER)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest'],\n content: `Source: Triple Whale / Klaviyo / Common Thread Collective MER convention.\n\n### Formula\n- MER = total_revenue / total_marketing_spend\n- Blended MER = ALL revenue (organic + paid + direct) over ALL paid spend. Headline number.\n- Channel MER = channel_attributed_revenue / channel_spend. Use last-click for simplicity.\n- nMER (new-customer MER) = new_customer_revenue / spend. More honest for acquisition.\n- aMER = first-order_revenue_from_new / spend.\n\n### Thresholds\n- Blended MER ≥ 3.0x = healthy. ≥ 5.0x = world-class. < 2.0x = unprofitable for most categories.\n- nMER ≥ 1.5x = acceptable at scale. ≥ 2.5x = strong.\n- Platform ROAS typically inflates by 2-3x vs MER. Trust MER for budget decisions.\n- Category: DTC apparel 2.5-3.5x, supplements 3.5-5x, B2B SaaS 4-8x.\n\n### When/Then\n- Channel MER < blended → dragging average down. Divest or restructure.\n- Channel MER > blended by > 50% → scale +20%, re-measure after 7 days.\n- Blended MER falls > 20% WoW → diagnose: spend up? revenue down? Both? Each implies different action.\n- nMER < 1 but blended > 3 → acquisition unprofitable; base carrying. Short-term only.\n- MER rises but new-customer count falls → harvesting existing demand. Collapses when base saturates.\n\n### Channel decision rule\nFor each channel with spend > $1k/week:\n- Score = (MER_c - blended) / blended\n- > +0.3 → +20% budget. < -0.2 → -50% or pause. Between → hold.\n\n### Anti-pattern\n- Quoting platform ROAS (Meta says 4x) as if equivalent to blended MER (1.8x). Platform ROAS double-counts.\n- 7-day MER window when sales cycle is 30d → lag artifacts. Use cycle-aligned window.\n- Excluding offline/email revenue → MER inflates (spend includes channels driving offline).\n- Counting subscription renewals in MER → renewals not caused by current spend. Use nMER for acquisition.\n\n### Interpretation for chat\n- \"Is paid working?\" → quote blended MER, nMER, and single best-delta channel. Always pair MER with trend.\n- MER alone insufficient. Pair with CAC trend + customer count to avoid the harvesting trap.\n- Use MER, not ROAS, in any cross-channel comparison.\n\n### Decision rules\n- Blended MER < 2.0x for 2 weeks → recommendations: \"Spend efficiency review.\"\n- Channel MER < 1.0x for 1 week → digest \"Loss-making channel: <name>.\"\n- nMER < 1.0x with blended ≥ 3.0x → digest \"Acquisition unprofitable; base carrying.\"\n- MER drop of > 30% week-over-week → standup headline regardless of absolute level.\n- For attribution-incremental teams: pair MER with a quarterly geo-lift or holdout test; MER alone over-credits brand/organic compounding to paid channels.`,\n },\n\n {\n name: 'Pipeline Coverage Ratio',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest', 'standup'],\n content: `Source: Sales-ops standard (Salesforce State of Sales, RevOps Co-op).\n\n### Formula\n- Coverage = open_pipeline_value / sales_quota_remaining (same period)\n- Pipeline = Σ deal_amount for stages discovery → proposal → negotiation. Exclude closed-won/lost.\n- Weighted coverage = Σ(deal × stage_probability) / quota. Weights: discovery 10%, qualified 25%, proposal 50%, negotiation 75%.\n- Forward coverage = next-quarter open pipeline / next-quarter quota. Early warning.\n\n### Thresholds\n- Healthy: ≥ 3.0x raw for next quarter (assumes ~33% win rate).\n- At-risk: 2.0-2.9x — likely to miss without intervention.\n- Off-track: < 2.0x — will miss barring outliers.\n- World-class: 4-5x at Q start, declining to 1.5-2x at Q end.\n- Win-rate-adjusted target = 1 / win_rate. 25% win rate → need 4x. 40% → 2.5x.\n\n### When/Then\n- Coverage < 2.0x with > 6 weeks left → trigger ToFu campaign (paid + outbound).\n- Coverage < 2.0x with < 4 weeks left → triage existing; chase commit-stage only. New pipeline won't close.\n- Coverage > 5x AND win rate trending down → bloated with low-quality deals. Pipeline hygiene sweep.\n- Coverage stable but avg deal size falling → ICP drift; review ICP.\n- Forward coverage (next Q) < 1.5x at current Q midpoint → escalate; pipeline build too slow.\n\n### Stage velocity check\n- Track median days_in_stage. If rises 30%+ vs trailing 90d → deals stuck. Coverage may be inflated by zombies.\n- Auto-flag deals with days_in_stage > 2× median for AE review.\n\n### Diagnostic decomposition\nShortfall = (quota - covered) / avg_deal_size = N new deals needed.\n- N > 10× current weekly new-deal rate → unrealistic; cut quota.\n- N achievable → recommend specific outbound sequences / paid bump.\n\n### Anti-pattern\n- Counting same deal across periods (rolling 12-mo pipeline vs quarterly quota) → inflates 3-4x.\n- Ignoring win-rate drift. Static 3x target wrong if win rate dropped 30%→20% — need 5x.\n- Treating all pipeline equally. Late-stage worth ~5x early-stage in real revenue.\n- Pumping low-quality deals to hit coverage → destroys win rate + forecast accuracy.\n- Unweighted as only number. Weighted = truer signal.\n\n### Interpretation for chat\n- \"Are we on track this quarter?\" → return weighted coverage, raw coverage, win rate, deals-needed-to-hit. All four.\n- Coverage requires quota. If unknown, ask once, then compute.\n- Pair coverage with stage-velocity sanity check; high coverage + slow velocity = false comfort.\n\n### Decision rules (Nexus)\n- Coverage < 2.0x → standup headline \"Pipeline gap: need $X more open by EOQ.\"\n- Coverage 2.0-2.9x → digest \"At-risk; ToFu push recommended.\"\n- Coverage trending down 2 weeks → recommendations: \"Outbound sequence boost\" + \"Paid ToFu spike.\"\n- Coverage > 5x + stage velocity down 30% → recommendations: \"Pipeline hygiene sweep.\"`,\n },\n\n {\n name: 'ICE Prioritization (Sean Ellis)',\n kind: 'framework',\n targets: ['ice_suggest', 'chat'],\n content: `Source: Sean Ellis, 2010 (LogMeIn, Dropbox growth team). For high-velocity experiment triage.\n\n### Formula\n- ICE = (Impact + Confidence + Ease) / 3\n- Each axis 1-10. Integer scoring fine; >1-decimal is false precision.\n- Impact: estimated NSM lift IF experiment wins. 10 = doubles NSM; 1 = barely measurable.\n- Confidence: P(experiment wins as hypothesized). 10 = replicates known win; 1 = wild guess.\n- Ease: inverse cost to ship+measure. 10 = afternoon, no eng; 1 = quarter-long.\n\n### Thresholds\n- ICE ≥ 8.0 → green-light immediately.\n- ICE 7.0-7.9 → next sprint. Default \"worth running\" cutoff.\n- ICE 5.0-6.9 → backlog. Re-score when context shifts.\n- ICE < 5.0 → reject or rework.\n\n### When/Then\n- 3+ experiments score ≥ 7 same week → parallel only if different funnel stages; else sequence (avoid interaction effects).\n- All cluster 5-7 → backlog stale; force brainstorming round on Impact-10 ideas.\n- Top-ICE all Ease=9 quick wins but NSM flat 4 weeks → optimizing locally. Force ≥1 high-Impact/low-Ease per sprint.\n- Confidence consistently > 8 → not exploring; bias toward more uncertain experiments.\n- Confidence consistently < 4 → guessing; invest in research before more experiments.\n\n### Anti-pattern\n- Chasing low-Ease wins that don't move NSM. Quick wins ≠ growth. Sean Ellis explicitly: ICE is for fast triage, NOT serious trade-off prioritization. Use RICE or WSJF for hard trade-offs.\n- Committee-average without dissent. Score independently, discuss outliers.\n- Anchoring Confidence on Impact. Score independently.\n- One person owning all three scores → consistent bias.\n\n### Variants\n- RICE = (Reach × Impact × Confidence) / Effort. Use when Reach varies widely.\n- WSJF = (User-Biz Value + Time Criticality + Risk Reduction) / Job Size. Cross-functional roadmap (SAFe).\n- Nexus stores all three. ICE = weekly growth triage; RICE = cross-segment audience; WSJF = roadmap vs eng capacity.\n\n### LLM scoring rules for ice_suggest\n- Impact from hypothesized NSM lift: +50% = 10, +25% = 8, +10% = 6, +5% = 4, +1% = 2, <1% = 1.\n- Confidence by similarity to historical wins in experiments table: direct replication = 9, similar pattern = 7, novel = 4, speculation = 2.\n- Ease from eng_effort_estimate_days: < 1d = 10, 1-3d = 8, 1w = 6, 2w = 4, > 1mo = 2.\n- Always show breakdown (I, C, E) + composite. Never return only the score.\n\n### Interpretation for chat\n- New experiment idea → score 1-10 per axis with one-line justification each.\n- Ranking a list → sort by ICE desc, tiebreak by Ease (cheaper first).\n- Refuse to score if Impact target metric unspecified. Ask \"lift on which metric?\" first.\n\n### Decision rules\n- ice_suggest auto-runs on any new hypothesis.\n- Top-3 surfaced in recommendations only if ICE ≥ 7 AND target metric matches an at-risk KPI from funnel_summary.`,\n },\n\n {\n name: 'AARRR Pirate Metrics',\n kind: 'framework',\n targets: ['funnel_summary', 'chat', 'correlations'],\n content: `Source: Dave McClure, 500 Startups, 2007. Five-stage user lifecycle for early-stage growth.\n\n### Stages\n1. Acquisition — user arrives. Metrics: sessions, signups, signup conv, CAC.\n2. Activation — first valuable experience (aha). Metrics: activation rate, time-to-activate, onboarding completion.\n3. Retention — user returns. Metrics: D1/D7/D30, WAU/MAU, churn.\n4. Referral — user invites. Metrics: viral coefficient K, invites/user, acceptance.\n5. Revenue — user pays. Metrics: ARPU, conv to paid, expansion, LTV.\n\n### Mapping to T-M-B funnel (Nexus)\n- TOFU = Acquisition + Activation\n- MOFU = Retention\n- BOFU = Referral + Revenue\nCompresses 5 stages into 3-tier display while preserving signal.\n\n### Stage health\n- Normalize each metric to [0, 100] vs industry benchmark:\n - higher better: score = clip(value / benchmark × 100, 0, 100)\n - lower better: score = clip(benchmark / value × 100, 0, 100)\n- Stage_health(s) = mean(normalized scores).\n\n### Thresholds\n- Stage health ≥ 75 = strong. 50-74 = OK. < 50 = bottleneck (flag).\n- AARRR weights (Nexus): A 0.20, A 0.20, R 0.25, R 0.15, R 0.20 (Retention emphasized).\n- T-M-B weights: TOFU 0.25, MOFU 0.40, BOFU 0.35.\n\n### When/Then\n- Single stage < 50 → THE bottleneck. Do NOT optimize others until fixed. Sequential.\n- Acquisition strong + Activation < 50 → leaky onboarding. Onboarding experiment.\n- Activation + Retention both < 50 → PMF issue, not marketing. Pause acquisition until fixed.\n- Retention strong + Revenue < 50 → monetization issue. Pricing/packaging review.\n- Referral < 50 + Revenue strong → leaving growth on table. Referral program experiment.\n- All 50-75 → balanced; pick biggest delta-to-benchmark for next experiment.\n\n### Correlation signals\n- Acquisition spike + Activation drop → traffic quality degraded (audience/creative mismatch).\n- Retention drop + Revenue stable → expansion masking churn.\n- Referral up + CAC down → viral loop firing; scale infra.\n- Activation up + Retention flat → fake activation event (definition too loose).\n\n### Anti-pattern\n- Optimizing acquisition before activation solid. Leaky bucket. McClure's original sin.\n- Picking activation event by ease-of-tracking, not aha-moment correlation with retention. Validate: users who hit X retain Z% vs Y% baseline.\n- AARRR as strict UX sequence — it's measurement framework, not journey map. Users can refer before activating.\n- Equal-weighting all stages. Retention compounds; weight higher.\n\n### Interpretation for chat\n- \"Where's the bottleneck?\" → stage health for all 5 (or T-M-B 3) with normalized scores. Identify single lowest, recommend that stage's top-ICE experiments.\n- Never propose Revenue optimization when Retention is the bottleneck. Sequence matters.\n- Use AARRR vocab in B2C/freemium. Use T-M-B/pipeline vocab in B2B sales-led.\n\n### Decision rules\n- funnel_summary returns stage_health(stage) + overall_health.\n- correlations engine scans stage-pair signals above and emits alerts.`,\n },\n\n {\n name: 'Funnel Health Scoring',\n kind: 'formula',\n targets: ['funnel_summary', 'recommendations', 'chat'],\n content: `Source: Nexus internal scoring (AARRR + GA4/HubSpot benchmarks). Used by api/intelligence.ts.\n\n### Formula\nFor each funnel stage s ∈ {TOFU, MOFU, BOFU}:\n- Stage_health(s) = mean(normalize(m_i, benchmark_i, direction_i)).\n- normalize(value, benchmark, direction):\n - 'up' (higher better): clip(value / benchmark × 100, 0, 100)\n - 'down' (lower better): clip(benchmark / value × 100, 0, 100)\n- Overall_health = 0.25 × TOFU + 0.40 × MOFU + 0.35 × BOFU.\n- Trend bonus: ±15 if delta_7d direction confirms benchmark improvement. Cap final at 100.\n\n### Metric → stage mapping (default)\n- TOFU: sessions, new_users, signup_rate, bounce_rate(↓), CPC(↓), impressions, CTR.\n- MOFU: activation_rate, D7_retention, MQL_count, lead_score_dist, time_to_activate(↓), trial_to_paid_rate.\n- BOFU: SQL_count, win_rate, avg_deal_size, MER, LTV/CAC, NRR, churn_rate(↓).\n\n### Benchmark anchors (B2B SaaS defaults)\nbounce_rate 50%, signup_conv 3%, activation_rate 35%, D7_retention 30%, trial_to_paid 15%, win_rate 25%, NRR 110%, monthly logo churn 2%, LTV/CAC 3.0.\n\n### Thresholds\n- Stage ≥ 75 = green. 50-74 = yellow. < 50 = red.\n- Overall ≥ 70 = healthy. 50-69 = mixed. < 50 = systemic issue.\n\n### When/Then\n- Stage < 50 → \"bottleneck\" alert; recommendations surfaces stage-specific experiments first.\n- Stage declines > 10 pts WoW → \"rapid decline\" alert even if absolute still > 50.\n- TOFU green + MOFU red → traffic quality OR onboarding. Audience refinement + activation experiment.\n- MOFU green + BOFU red → conversion/monetization. Pricing or sales-process experiment.\n- All green but overall flat → optimization plateau; new-channel exploration.\n- Trend bonus pushes 49→64 → label \"improving\" but use RAW score as action signal.\n\n### Trend bonus\n- delta_7d = today - 7d_ago. bonus = clip(delta_7d × 0.5, -15, +15). Reported = clip(raw + bonus, 0, 100).\n- Recommendations engine uses RAW for thresholds. Bonus is UI/digest only.\n\n### Decomposition\nWhen stage drops: delta_contribution(m_i) = (norm_today - norm_7d) × (1 / N_metrics). Surface top-2 negative contributors in digest.\n\n### Anti-pattern\n- Equal weighting across stages. Retention/BOFU compound; weight higher (0.40 MOFU, 0.35 BOFU).\n- Single-metric stage health (only bounce_rate for TOFU). ≥3 metrics per stage to reduce noise.\n- Single industry benchmark for multi-segment business. Use segment-specific when SMB/Ent mix > 30/70.\n- Confusing trend bonus with action signal. Bonus = narrative; raw = decisions.\n- Bench-marking against world-class for young companies — perpetual red flags.\n\n### Interpretation for chat\n- \"How's the funnel?\" → TOFU/MOFU/BOFU health (raw + reported), overall, top-2 declining metrics.\n- Always cite benchmark used. e.g. \"TOFU 62/100 because bounce_rate 58% vs benchmark 50%.\"\n- Overall > 70 but single stage < 50 → lead with bottleneck, not headline.\n\n### Decision rules\n- raw stage_health < 50 → focus for recommendations top-3.\n- delta_7d_health < -10 → push to standup regardless of absolute level.`,\n },\n\n {\n name: 'MQL/SQL Lead Scoring',\n kind: 'framework',\n targets: ['lead_scoring', 'chat', 'digest'],\n content: `Source: HubSpot / Marketo conventions, adapted for Nexus signals table.\n\n### MQL score formula\nSum activity points over rolling 30d window with decay:\n- +5 email open / passive page view\n- +15 link click (email or content)\n- +25 pricing / plans / cost mention or page visit\n- +20 gated content download (whitepaper, ebook, calculator)\n- +20 webinar registration or attendance\n- +10 repeat email open within 7d (max 3× = +30)\n- +10 return visit within 7d\n\n### SQL signals (boolean count)\n- Reply to outreach = 1; Demo request = 1; \"Talk to sales\" form = 1; Calendar event booked = 1; Direct inbound call = 1.\n\n### Qualification rules\n- mql_qualified = score ≥ 40 in 30-day window.\n- sql_qualified = sql_signal_count ≥ 2 OR any single signal in {demo_request, calendar_event} (single-signal qualifies — see intent_signal_weights).\n- Hot lead (outreach priority): score ≥ 75.\n- Cold (deprioritize): score < 15 AND no signals in 30d.\n\n### Decay\n- Signals > 14d old: × 0.5.\n- Signals > 30d old: × 0.25 (effectively drop).\n- Apply at query time; do not destroy raw signal records.\n\n### Account-level bonus (B2B)\n- +15 if ≥ 3 signals from same domain within 7d. Captures buying-committee activity.\n- +10 if second person from domain registers a signal within 14d.\n- Account score = max(individual_score) + bonus, surfaced separately.\n\n### When/Then\n- Score ≥ 40 + not contacted → MQL queue; SDR SLA 24h.\n- Score ≥ 75 (hot) → AE queue; outreach SLA 4h.\n- sql_qualified AND no meeting in 7d → escalate to RevOps; coverage gap.\n- Score plateaus 40-60 for > 21d → nurture sequence; not sales-ready.\n- Score spikes < 20 → > 60 in 7d → \"fast riser\" tag; priority outreach.\n- demo_request from low-score lead (< 20) → trust the demo; demo > behavioral score.\n\n### Anti-pattern\n- No decay → stale leads pollute MQL queue.\n- Single-metric scoring (only opens). Easily gamed.\n- MQL volume as KPI without MQL→SQL conversion. Volume without quality = wasted SDR time.\n- Manual overrides without logging. Kills model improvement loop.\n- Same rules for inbound vs outbound — outbound needs different weights.\n\n### Interpretation for chat\n- \"Is lead X hot?\" → score, signal breakdown by category, recency. Include time-since-last-signal.\n- MQL queue → sort by score desc, but surface \"fast risers\" first regardless of absolute score.\n- Never qualify a lead without signal data; if signals_loaded=false, ask to refresh integrations.\n\n### Decision rules (Nexus)\n- lead_scoring engine recomputes daily; top-25 by score surfaced.\n- digest \"Hot leads this week\" = score ≥ 75 OR sql_qualified=true.\n- standup \"New SQLs overnight\" = transitions to sql_qualified=true in last 24h.\n\n### Conversion benchmarks\n- MQL→SQL: 13% B2B SaaS (Forrester). < 8% = quality issue. > 25% = threshold too high.\n- SQL→Won: 20-30%. < 15% = qualification issue (false positives).`,\n },\n\n {\n name: 'Intent Signal Weights (B2B SaaS)',\n kind: 'benchmark',\n targets: ['lead_scoring', 'digest'],\n content: `Source: Nexus signal weights, calibrated from HubSpot/6sense intent conventions.\n\n### Weight table (signal | pts | tier)\n- email_open 5 passive | email_open_repeat (7d) +5 ea cap +15 passive | email_click 15 engaged | email_reply 25 high-intent\n- page_view (non-pricing) 3 passive | pricing_page_view 25 high-intent | pricing_keyword 30 high-intent\n- content_download 20 engaged | webinar_registration 20 engaged | webinar_attendance 25 high-intent\n- demo_request 50 SQL-tier | calendar_event (booked) 50 SQL-tier\n- linkedin_view (company) 10 engaged | linkedin_engagement 15 engaged | g2_review_view (category) 15 engaged\n- competitor_comparison_visit 30 high-intent | pricing_calculator_use 35 high-intent | contract_template_download 35 high-intent | support_ticket_pre_sale 20 engaged\n\n### Tier rules\n- passive: alone insufficient; ≥ 3 needed.\n- engaged: contributes meaningfully; doesn't qualify alone.\n- high-intent: 2+ in 14d = MQL+; surface to SDR.\n- SQL-tier: SINGLE demo_request OR calendar_event = SQL.\n\n### Aggregation\n- Per lead: sum points over rolling 30d with decay.\n- Per account: max(lead_score) + account_bonus.\n- Buckets: 0-7d full, 7-14d ×0.5, 14-30d ×0.25, > 30d drop.\n\n### When/Then\n- demo_request OR calendar_event fires → auto-create SQL; sales SLA 4h.\n- pricing_page_view + competitor_comparison_visit within 7d same lead → in-market; assign AE now.\n- pricing_calculator_use → very strong intent; surface in standup \"high-intent activity overnight.\"\n- email_open_repeat hits cap (+15) but no clicks → interested in subject lines, not content. Refresh creative.\n- linkedin_view spikes from account with no other signals → research stage; tag, don't outreach yet.\n\n### Industry calibration\n- Enterprise (ACV > $50k): calendar_event 50→35; contract_template_download 35→50.\n- SMB / PLG (ACV < $5k): pricing_calculator_use 35→50; demo_request 50→30.\n- Do NOT use single weight table across heterogeneous segments.\n\n### Combination signals\n- pricing_page_view + competitor_comparison_visit within 7d → +25 (research depth).\n- demo_request + content_download within 14d prior → \"primed buyer\"; expedite.\n- email_reply + linkedin_engagement same person → multi-channel; high-quality lead.\n\n### Anti-pattern\n- Counting page_view as engaged. Drive-by traffic; keep passive (3 pts).\n- Equal-weighting demo_request and email_open. Off by 10x in actual intent.\n- Ignoring company-level signal (leads as independent). B2B is buying-committee.\n- Static weights for years. Recalibrate quarterly against actual SQL→Close conversion.\n\n### Interpretation for chat\n- List lead's signals in TIER order (SQL > high-intent > engaged > passive), not chronological.\n- \"Should I reach out?\" → any SQL-tier in 7d = yes now. High-intent stack ≥ 2 in 14d = yes today. Else nurture.\n- Always cite specific signals driving the score; opacity destroys trust.`,\n },\n\n {\n name: 'Engagement Velocity',\n kind: 'formula',\n targets: ['lead_scoring', 'digest', 'standup'],\n content: `Source: SaaS sales-cycle analytics (Bridge Group, Salesforce velocity reports).\n\n### Formula\n- velocity = mean(SQL_qualified_at - MQL_qualified_at) across leads transitioning in a period.\n- Use MEDIAN for skew-resistance; report both mean and median.\n- Compute per segment (SMB / mid / enterprise) — blended hides segment differences.\n\n### Variants\n- mql_to_sql_velocity = mean(SQL_at - MQL_at) — primary metric.\n- sql_to_close_velocity = mean(closed_won_at - SQL_at).\n- first_touch_to_close = mean(closed_won_at - first_signal_at) — full cycle.\n- mql_to_close = mql_to_sql + sql_to_close.\n\n### Benchmarks (B2B SaaS, ACV $5k-$50k)\n- mql_to_sql_velocity median: 14-21 days. < 14d = fast. > 30d = slow.\n- sql_to_close_velocity median: 30-60 days. Varies sharply with ACV.\n- Total cycle: 60-120 days SMB; 6-12 months enterprise.\n- PLG / freemium: mql_to_sql can be < 7 days if product usage drives qualification.\n\n### Thresholds\n- velocity < 7d → review qualification rigor; likely false positives.\n- velocity 7-14d → fast + healthy for SMB/mid-market.\n- velocity 14-30d → standard; benchmark range.\n- velocity > 30d → AE engagement quality or fit issue. Investigate.\n- velocity > 60d → process broken or wrong ICP being scored.\n\n### When/Then\n- velocity > 30d AND MQL→SQL < 13% → AE engagement quality issue. SDR coaching, faster touch SLA.\n- velocity > 30d AND MQL→SQL > 20% → leads qualify slowly. ICP fit good, motion slow. Shorter discovery, multi-thread sooner.\n- velocity < 7d AND MQL→SQL > 40% → qualification too loose. Raise sql_signals 2→3; audit recent SQLs.\n- velocity declines WoW → process improving OR threshold loosening. Pair with conversion to disambiguate.\n- velocity spikes up → likely SDR capacity / vacation. Check headcount before assuming process failure.\n\n### Decomposition\nvelocity = Σ stage_dwell: mql_to_first_touch + first_touch_to_discovery + discovery_to_sql. Surface LONGEST sub-stage as bottleneck.\n\n### Cohort version\nPer MQL_at week-cohort. Trailing 8-week table shows trend. Recent 2 cohorts > 30d while prior were 18d → regression; investigate last 2 weeks of process changes.\n\n### Anti-pattern\n- Averaging across segments. SMB 10d + Enterprise 90d blended = 50d, useless.\n- Counting only closed leads → excludes slow/stalled, biases low. Include all SQL-qualified, censor at 90d.\n- Minimize-as-target without watching conversion. Faster ≠ better if quality drops.\n- Computing weekly with n < 20. Use 4-week rolling window.\n\n### Interpretation for chat\n- \"Are leads moving fast enough?\" → mql_to_sql median vs benchmark, with conversion context. Two numbers, not one.\n- velocity changes > 20% WoW → standup highlight with cohort decomposition.\n- Refuse single velocity number if segment mix unknown; ask for segment.\n\n### Pipeline velocity (revenue formulation)\npipeline_velocity = (num_opps × win_rate × avg_deal_size) / sales_cycle_days. Dollars-per-day through pipeline. Use for revenue forecasting; engagement velocity (above) is for lead-flow diagnosis.\n\n### Decision rules (Nexus)\n- velocity > 30d for 2 consecutive 4-week windows → recommendations: \"AE engagement quality review.\"\n- velocity < 7d AND mql→sql > 40% → digest \"Qualification rigor: false-positive risk.\"\n- standup \"Yesterday's SQLs\" includes individual transition times; outliers (> 2× median) flagged.`,\n },\n];\n",[],[365,778],"5953099c-5b19-11f1-98e7-ee650bb23af1","llm.ts","31190010-56ca-11f1-8403-ee650bb23af1",468,["D",1780029985598],34,"import { sqlite } from \"../db.ts\";\nimport { isSafePublicUrl } from \"./integrations/_shared.ts\";\nimport type { LLMConfig, ChatMessage } from \"../types.ts\";\n\n// --- Defaults -----------------------------------------------------------------\n\nconst DEFAULT_CONFIG: LLMConfig = {\n id: \"default\",\n provider: \"anthropic\",\n model: \"claude-sonnet-4-6\",\n api_key: \"\",\n endpoint: \"\",\n};\n\n// Provider - default model. Used to auto-correct on provider switch so users don't\n// accidentally send \"claude-sonnet-4-6\" to Gemini (which 400s).\nconst PROVIDER_DEFAULT_MODEL: Record<string, string> = {\n anthropic: \"claude-sonnet-4-6\",\n openai: \"gpt-4o\",\n gemini: \"gemini-3-pro-preview\",\n custom: \"your-model\",\n};\nconst VALID_PROVIDERS = new Set([\"anthropic\", \"openai\", \"gemini\", \"custom\"]);\n\n// Map a provider name to its dedicated key column on llm_config.\nconst providerKeyCol = (p: string): string => {\n switch (p) {\n case \"anthropic\": return \"api_key_anthropic\";\n case \"openai\": return \"api_key_openai\";\n case \"gemini\": return \"api_key_gemini\";\n case \"custom\": return \"api_key_custom\";\n default: return \"api_key_anthropic\";\n }\n};\n\n// --- Helpers ------------------------------------------------------------------\n\nfunction rowToConfig(columns: string[], row: any[]): LLMConfig & { has_key?: Record<string, boolean> } {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n const provider = obj.provider ?? DEFAULT_CONFIG.provider;\n // Resolve api_key: prefer the per-provider column, fall back to legacy single api_key.\n const perProviderKey = obj[providerKeyCol(provider)];\n const api_key = (perProviderKey && String(perProviderKey)) || obj.api_key || \"\";\n const endpoint = provider === \"custom\" ? (obj.endpoint_custom || obj.endpoint || \"\") : (obj.endpoint || \"\");\n return {\n id: obj.id ?? \"default\",\n provider,\n model: obj.model ?? DEFAULT_CONFIG.model,\n api_key,\n endpoint,\n has_key: {\n anthropic: !!(obj.api_key_anthropic || (provider === \"anthropic\" && obj.api_key)),\n openai: !!(obj.api_key_openai || (provider === \"openai\" && obj.api_key)),\n gemini: !!(obj.api_key_gemini || (provider === \"gemini\" && obj.api_key)),\n custom: !!(obj.api_key_custom || (provider === \"custom\" && obj.api_key)),\n },\n } as any;\n}\n\n// --- DB operations ------------------------------------------------------------\n\nexport async function getLLMConfig(): Promise<LLMConfig> {\n const result = await sqlite.execute(\n `SELECT id, provider, model, api_key, endpoint,\n api_key_anthropic, api_key_openai, api_key_gemini, api_key_custom, endpoint_custom\n FROM llm_config WHERE id = 'default'`\n );\n\n if (!result.rows || result.rows.length === 0) {\n return { ...DEFAULT_CONFIG };\n }\n\n return rowToConfig(result.columns, result.rows[0]);\n}\n\nexport async function saveLLMConfig(\n config: Partial<LLMConfig>\n): Promise<LLMConfig> {\n const existing = await getLLMConfig();\n\n // Validate provider; fall back to existing if caller sent garbage.\n let provider = (config.provider ?? existing.provider) as string;\n if (!VALID_PROVIDERS.has(provider)) provider = existing.provider;\n\n // Auto-correct model when provider changes and caller didn't pick a sensible one.\n // Prevents \"model not found\" 400s when switching anthropic - gemini.\n let model = config.model ?? existing.model;\n if (provider !== existing.provider) {\n const userPickedNewModel = config.model && config.model !== existing.model;\n if (!userPickedNewModel) {\n model = PROVIDER_DEFAULT_MODEL[provider] ?? existing.model;\n }\n }\n\n // Per-provider key: if caller supplied api_key, route it to that provider's column.\n // Other providers' keys are preserved across switches.\n const incomingKey = config.api_key;\n const keyCol = providerKeyCol(provider);\n const endpoint = provider === \"custom\" ? (config.endpoint ?? existing.endpoint ?? \"\") : (existing.endpoint ?? \"\");\n\n const now = new Date().toISOString();\n // Upsert: keep existing per-provider columns, only overwrite the one(s) the user changed.\n await sqlite.execute(\n `INSERT INTO llm_config (id, provider, model, api_key, endpoint, ${keyCol}, endpoint_custom, updated_at)\n VALUES ('default', ?, ?, ?, ?, ?, ?, ?)\n ON CONFLICT(id) DO UPDATE SET\n provider = excluded.provider,\n model = excluded.model,\n api_key = excluded.api_key,\n endpoint = excluded.endpoint,\n ${keyCol} = COALESCE(NULLIF(excluded.${keyCol}, ''), llm_config.${keyCol}),\n endpoint_custom = COALESCE(NULLIF(excluded.endpoint_custom, ''), llm_config.endpoint_custom),\n updated_at = excluded.updated_at`,\n [\n provider,\n model,\n incomingKey ?? existing.api_key ?? \"\",\n endpoint,\n incomingKey ?? \"\",\n provider === \"custom\" ? endpoint : \"\",\n now,\n ]\n );\n\n return await getLLMConfig();\n}\n\n// --- LLM call -----------------------------------------------------------------\n\nexport async function callLLM(\n config: LLMConfig,\n messages: ChatMessage[],\n options?: {\n systemPrompt?: string;\n tools?: any[];\n maxTokens?: number;\n temperature?: number; // 0..1 - passed to provider\n frequencyPenalty?: number; // 0..2 - OpenAI/custom only (Anthropic ignores)\n }\n): Promise<{ content: string; tool_calls?: any[] }> {\n const maxTokens = options?.maxTokens ?? 4096;\n const systemPrompt = options?.systemPrompt;\n const tools = options?.tools;\n const temperature = options?.temperature;\n const frequencyPenalty = options?.frequencyPenalty;\n\n if (config.provider === \"anthropic\") {\n return callAnthropic(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"openai\") {\n return callOpenAI(config, \"https://api.openai.com/v1\", messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else if (config.provider === \"gemini\") {\n return callGemini(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"custom\") {\n const endpoint = config.endpoint?.replace(/\\/+$/, \"\") ?? \"\";\n // SSRF guard: a custom LLM endpoint is user-supplied and fetched server-side with the API key.\n if (!isSafePublicUrl(endpoint)) {\n throw new Error(\"Custom LLM endpoint must be a public https:// URL\");\n }\n return callOpenAI(config, endpoint, messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else {\n throw new Error(`Unknown LLM provider: ${config.provider}`);\n }\n}\n\n// --- Streaming LLM call -------------------------------------------------------\n\nexport async function* callLLMStream(\n config: LLMConfig,\n messages: ChatMessage[],\n options?: {\n systemPrompt?: string;\n tools?: any[];\n maxTokens?: number;\n temperature?: number;\n frequencyPenalty?: number;\n }\n): AsyncGenerator<string> {\n const maxTokens = options?.maxTokens ?? 4096;\n const systemPrompt = options?.systemPrompt;\n const tools = options?.tools;\n const temperature = options?.temperature;\n const frequencyPenalty = options?.frequencyPenalty;\n\n if (config.provider === \"anthropic\") {\n yield* callAnthropicStream(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"openai\") {\n yield* callOpenAIStream(config, \"https://api.openai.com/v1\", messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else if (config.provider === \"gemini\") {\n yield* callGeminiStream(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"custom\") {\n const endpoint = config.endpoint?.replace(/\\/+$/, \"\") ?? \"\";\n if (!isSafePublicUrl(endpoint)) {\n throw new Error(\"Custom LLM endpoint must be a public https:// URL\");\n }\n yield* callOpenAIStream(config, endpoint, messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else {\n throw new Error(`Unknown LLM provider: ${config.provider}`);\n }\n}\n\nasync function* parseSseStream(body: ReadableStream<Uint8Array>, onChunk: (parsed: any) => string | null): AsyncGenerator<string> {\n const reader = body.getReader();\n const dec = new TextDecoder();\n let buf = \"\";\n const chunks: string[] = [];\n\n const flush = async function* () {\n while (chunks.length) {\n yield chunks.shift()!;\n }\n };\n\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) break;\n buf += dec.decode(value, { stream: true });\n const lines = buf.split(\"\\n\");\n buf = lines.pop()!;\n for (const line of lines) {\n if (!line.startsWith(\"data: \")) continue;\n const d = line.slice(6).trim();\n if (d === \"[DONE]\") return;\n try {\n const j = JSON.parse(d);\n const text = onChunk(j);\n if (text) yield text;\n } catch { /* skip malformed SSE lines */ }\n }\n }\n } finally {\n reader.cancel().catch(() => {});\n }\n}\n\nasync function* callAnthropicStream(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): AsyncGenerator<string> {\n const body: Record<string, any> = {\n model: config.model ?? \"claude-sonnet-4-6\",\n max_tokens: maxTokens,\n messages: toAnthropicMessages(messages),\n stream: true,\n };\n if (typeof temperature === \"number\") body.temperature = Math.max(0, Math.min(1, temperature));\n if (systemPrompt) body.system = systemPrompt;\n if (tools && tools.length > 0) {\n body.tools = tools.map((t) => {\n if (t.type === \"function\") {\n return { name: t.function.name, description: t.function.description ?? \"\", input_schema: t.function.parameters ?? { type: \"object\", properties: {} } };\n }\n return t;\n });\n }\n\n const res = await fetch(\"https://api.anthropic.com/v1/messages\", {\n method: \"POST\",\n headers: { \"x-api-key\": config.api_key, \"anthropic-version\": \"2023-06-01\", \"Content-Type\": \"application/json\" },\n body: JSON.stringify(body),\n });\n if (!res.ok) throw new Error(`Anthropic ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n if (j.type === \"content_block_delta\" && j.delta?.text) return j.delta.text;\n return null;\n });\n}\n\nasync function* callOpenAIStream(\n config: LLMConfig,\n baseUrl: string,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number,\n frequencyPenalty?: number\n): AsyncGenerator<string> {\n const openAIMessages: any[] = [];\n if (systemPrompt) openAIMessages.push({ role: \"system\", content: systemPrompt });\n for (const m of messages as any[]) {\n if (m.role === \"assistant\" && m.tool_calls) {\n const msg: any = { role: \"assistant\", content: m.content ?? null, tool_calls: m.tool_calls };\n if (m.reasoning_content) msg.reasoning_content = m.reasoning_content;\n openAIMessages.push(msg);\n } else if (m.role === \"tool\") {\n openAIMessages.push({ role: \"tool\", tool_call_id: m.tool_call_id, content: String(m.content ?? \"\") });\n } else {\n openAIMessages.push({ role: m.role, content: m.content ?? \"\" });\n }\n }\n\n const body: Record<string, any> = {\n model: config.model ?? \"gpt-4o\",\n messages: openAIMessages,\n max_tokens: maxTokens,\n temperature: typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 0.7,\n stream: true,\n };\n if (typeof frequencyPenalty === \"number\") body.frequency_penalty = Math.max(0, Math.min(2, frequencyPenalty));\n if (tools && tools.length > 0) { body.tools = tools; body.tool_choice = \"auto\"; }\n\n const res = await fetch(`${baseUrl}/chat/completions`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${config.api_key}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify(body),\n });\n if (!res.ok) throw new Error(`OpenAI ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n const text = j.choices?.[0]?.delta?.content;\n if (text) return text;\n return null;\n });\n}\n\nasync function* callGeminiStream(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): AsyncGenerator<string> {\n const model = config.model ?? \"gemini-3-pro-preview\";\n // Use ?alt=sse for server-sent events streaming\n const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:streamGenerateContent?alt=sse`;\n\n const tcNameMap: Record<string, string> = {};\n for (const m of messages as any[]) {\n if (m.tool_calls) for (const tc of m.tool_calls) tcNameMap[tc.id] = tc.function.name;\n }\n\n const contents: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"system\") continue;\n if (m.role === \"tool\") {\n const fnName = tcNameMap[m.tool_call_id] ?? m.tool_call_id ?? \"tool\";\n const part = { functionResponse: { id: (m as any).tool_call_id, name: fnName, response: { content: m.content } } };\n const last = contents[contents.length - 1];\n if (last?.role === \"user\") last.parts.push(part);\n else contents.push({ role: \"user\", parts: [part] });\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const parts: any[] = [];\n if (m.content) parts.push({ text: m.content });\n for (const tc of m.tool_calls) {\n let args: any = {};\n try { args = typeof tc.function.arguments === \"string\" ? JSON.parse(tc.function.arguments) : (tc.function.arguments ?? {}); } catch { args = {}; }\n parts.push({ functionCall: { id: tc.id, name: tc.function.name, args } });\n }\n contents.push({ role: \"model\", parts });\n } else {\n const role = m.role === \"assistant\" ? \"model\" : \"user\";\n contents.push({ role, parts: [{ text: m.content }] });\n }\n }\n\n const genConfig: Record<string, any> = { maxOutputTokens: maxTokens };\n genConfig.temperature = typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 1.0;\n genConfig.thinkingConfig = { thinkingLevel: \"medium\" };\n\n const reqBody: Record<string, any> = { contents, generationConfig: genConfig };\n if (systemPrompt) reqBody.systemInstruction = { parts: [{ text: systemPrompt }] };\n if (tools && tools.length > 0) {\n reqBody.tools = [{ functionDeclarations: tools.map(t => { const fn = t.type === \"function\" ? t.function : t; return { name: fn.name, description: fn.description ?? \"\", parameters: stripUnsupported(fn.parameters ?? { type: \"object\", properties: {} }) }; }) }];\n reqBody.toolConfig = { functionCallingConfig: { mode: \"AUTO\" } };\n }\n\n const res = await fetch(url, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\", \"x-goog-api-key\": config.api_key },\n body: JSON.stringify(reqBody),\n });\n if (!res.ok) throw new Error(`Gemini ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n const text = j.candidates?.[0]?.content?.parts?.[0]?.text;\n if (text) return text;\n return null;\n });\n}\n\n// --- Anthropic ----------------------------------------------------------------\n\nfunction toAnthropicMessages(messages: ChatMessage[]): any[] {\n const result: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"tool\") {\n const block = { type: \"tool_result\", tool_use_id: m.tool_call_id, content: m.content };\n const prev = result[result.length - 1];\n if (prev && prev.role === \"user\" && Array.isArray(prev.content)) {\n prev.content.push(block);\n } else {\n result.push({ role: \"user\", content: [block] });\n }\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const content: any[] = [];\n if (m.content) content.push({ type: \"text\", text: m.content });\n for (const tc of m.tool_calls) {\n let input: any = {};\n try {\n input = typeof tc.function.arguments === \"string\"\n ? JSON.parse(tc.function.arguments) : tc.function.arguments ?? {};\n } catch { input = {}; }\n content.push({ type: \"tool_use\", id: tc.id, name: tc.function.name, input });\n }\n result.push({ role: \"assistant\", content });\n } else {\n result.push({ role: m.role, content: m.content });\n }\n }\n return result;\n}\n\nasync function callAnthropic(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n const body: Record<string, any> = {\n model: config.model ?? \"claude-sonnet-4-6\",\n max_tokens: maxTokens,\n messages: toAnthropicMessages(messages),\n };\n if (typeof temperature === \"number\") body.temperature = Math.max(0, Math.min(1, temperature));\n\n if (systemPrompt) {\n body.system = systemPrompt;\n }\n\n if (tools && tools.length > 0) {\n // Convert OpenAI-style tools to Anthropic format if needed\n body.tools = tools.map((t) => {\n if (t.type === \"function\") {\n return {\n name: t.function.name,\n description: t.function.description ?? \"\",\n input_schema: t.function.parameters ?? { type: \"object\", properties: {} },\n };\n }\n // Already in Anthropic format\n return t;\n });\n }\n\n // 25s timeout per LLM call (see callOpenAI for rationale).\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(\"https://api.anthropic.com/v1/messages\", {\n method: \"POST\",\n headers: {\n \"x-api-key\": config.api_key,\n \"anthropic-version\": \"2023-06-01\",\n \"Content-Type\": \"application/json\",\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`Anthropic API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n\n // Extract text content and tool use blocks\n let textContent = \"\";\n const toolCalls: any[] = [];\n\n if (Array.isArray(data.content)) {\n for (const block of data.content) {\n if (block.type === \"text\") {\n textContent += block.text;\n } else if (block.type === \"tool_use\") {\n toolCalls.push({\n id: block.id,\n type: \"function\",\n function: {\n name: block.name,\n arguments: JSON.stringify(block.input),\n },\n });\n }\n }\n }\n\n return {\n content: textContent,\n tool_calls: toolCalls.length > 0 ? toolCalls : undefined,\n };\n}\n\n// --- OpenAI / OpenAI-compatible ----------------------------------------------\n\nasync function callOpenAI(\n config: LLMConfig,\n baseUrl: string,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number,\n frequencyPenalty?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n const openAIMessages: any[] = [];\n\n if (systemPrompt) {\n openAIMessages.push({ role: \"system\", content: systemPrompt });\n }\n\n for (const m of messages as any[]) {\n if (m.role === \"assistant\" && m.tool_calls) {\n const msg: any = { role: \"assistant\", content: m.content ?? null, tool_calls: m.tool_calls };\n // DeepSeek thinking mode: reasoning_content must be echoed back for multi-turn\n if (m.reasoning_content) msg.reasoning_content = m.reasoning_content;\n openAIMessages.push(msg);\n } else if (m.role === \"tool\") {\n openAIMessages.push({ role: \"tool\", tool_call_id: m.tool_call_id, content: String(m.content ?? \"\") });\n } else {\n openAIMessages.push({ role: m.role, content: m.content ?? \"\" });\n }\n }\n\n const body: Record<string, any> = {\n model: config.model ?? \"gpt-4o\",\n messages: openAIMessages,\n max_tokens: maxTokens,\n temperature: typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 0.7,\n };\n if (typeof frequencyPenalty === \"number\") body.frequency_penalty = Math.max(0, Math.min(2, frequencyPenalty));\n\n if (tools && tools.length > 0) {\n body.tools = tools;\n body.tool_choice = \"auto\";\n }\n\n // Hard 25s timeout per LLM call. Without this, a slow/unreachable provider hangs\n // the whole chat handler past Val Town's 60s execution cap and triggers a pause.\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(`${baseUrl}/chat/completions`, {\n method: \"POST\",\n headers: {\n Authorization: `Bearer ${config.api_key}`,\n \"Content-Type\": \"application/json\",\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`OpenAI API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n const choice = data.choices?.[0];\n const message = choice?.message;\n\n const content: string = message?.content ?? \"\";\n const tool_calls: any[] | undefined =\n message?.tool_calls && message.tool_calls.length > 0\n ? message.tool_calls\n : undefined;\n // DeepSeek thinking mode: preserve reasoning_content so it can be echoed back in multi-turn\n const reasoning_content: string | undefined = message?.reasoning_content ?? undefined;\n\n return { content, tool_calls, reasoning_content };\n}\n\n// --- Gemini (native REST API) -------------------------------------------------\n\n// Gemini rejects additionalProperties and other JSON Schema fields it doesn't know\nfunction stripUnsupported(schema: any): any {\n if (!schema || typeof schema !== \"object\") return schema;\n const { additionalProperties, $schema, ...rest } = schema;\n if (rest.properties) {\n rest.properties = Object.fromEntries(\n Object.entries(rest.properties).map(([k, v]) => [k, stripUnsupported(v)])\n );\n }\n if (rest.items) rest.items = stripUnsupported(rest.items);\n return rest;\n}\n\nasync function callGemini(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n // Gemini 3 migration: default model is now gemini-3-pro-preview (gemini-1.5-flash deprecated).\n // Auth moved from ?key= query to the x-goog-api-key header (canonical).\n const model = config.model ?? \"gemini-3-pro-preview\";\n const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:generateContent`;\n\n // Build tool_call_id -> function name map so functionResponse can use the name\n const tcNameMap: Record<string, string> = {};\n for (const m of messages as any[]) {\n if (m.tool_calls) {\n for (const tc of m.tool_calls) {\n tcNameMap[tc.id] = tc.function.name;\n }\n }\n }\n\n const contents: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"system\") continue;\n\n if (m.role === \"tool\") {\n const fnName = tcNameMap[m.tool_call_id] ?? m.tool_call_id ?? \"tool\";\n // id must match the corresponding FunctionCall id (required in Gemini 3.x)\n const part = { functionResponse: { id: (m as any).tool_call_id, name: fnName, response: { content: m.content } } };\n const last = contents[contents.length - 1];\n if (last?.role === \"user\") {\n last.parts.push(part);\n } else {\n contents.push({ role: \"user\", parts: [part] });\n }\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const parts: any[] = [];\n if (m.content) parts.push({ text: m.content });\n for (const tc of m.tool_calls) {\n let args: any = {};\n try { args = typeof tc.function.arguments === \"string\" ? JSON.parse(tc.function.arguments) : (tc.function.arguments ?? {}); } catch { args = {}; }\n const fcPart: any = { functionCall: { id: tc.id, name: tc.function.name, args } };\n // Echo thoughtSignature exactly as received - required by Gemini 2.x thinking models\n if (tc._gemini_thought_signature) fcPart.thoughtSignature = tc._gemini_thought_signature;\n parts.push(fcPart);\n }\n contents.push({ role: \"model\", parts });\n } else {\n const role = m.role === \"assistant\" ? \"model\" : \"user\";\n contents.push({ role, parts: [{ text: m.content }] });\n }\n }\n\n // Gemini 3 defaults: temperature 1.0 (research confirms lowering it degrades reasoning).\n // We honor an explicit caller override only when in [0,2] (Gemini 3 expanded range).\n const genConfig: Record<string, any> = { maxOutputTokens: maxTokens };\n genConfig.temperature = typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 1.0;\n // Gemini 3 introduced thinkingLevel (\"minimal\"|\"low\"|\"medium\"|\"high\") replacing the\n // numeric thinking_budget. We default to \"medium\" - Gemini 3's reasoning mode.\n genConfig.thinkingConfig = { thinkingLevel: \"medium\" };\n\n const body: Record<string, any> = {\n contents,\n generationConfig: genConfig,\n };\n\n if (systemPrompt) {\n body.systemInstruction = { parts: [{ text: systemPrompt }] };\n }\n\n if (tools && tools.length > 0) {\n body.tools = [{\n functionDeclarations: tools.map(t => {\n const fn = t.type === \"function\" ? t.function : t;\n return { name: fn.name, description: fn.description ?? \"\", parameters: stripUnsupported(fn.parameters ?? { type: \"object\", properties: {} }) };\n }),\n }];\n body.toolConfig = { functionCallingConfig: { mode: \"AUTO\" } };\n }\n\n // 25s timeout per LLM call (see callOpenAI for rationale).\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(url, {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n // Gemini 3 canonical auth - header replaces the legacy ?key= query param.\n \"x-goog-api-key\": config.api_key,\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`Gemini API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n const parts = data.candidates?.[0]?.content?.parts ?? [];\n\n let textContent = \"\";\n const toolCalls: any[] = [];\n\n for (const part of parts) {\n if (part.text) textContent += part.text;\n else if (part.functionCall) {\n // Use the model's own id - must be echoed exactly in the functionResponse\n const tc: any = {\n id: part.functionCall.id ?? `gemini_${Date.now()}_${toolCalls.length}`,\n type: \"function\",\n function: {\n name: part.functionCall.name,\n arguments: JSON.stringify(part.functionCall.args ?? {}),\n },\n };\n // Preserve thoughtSignature (Part-level field) - required by Gemini 2.x thinking models\n const sig = part.thoughtSignature ?? part.thought_signature;\n if (sig) tc._gemini_thought_signature = sig;\n toolCalls.push(tc);\n }\n }\n\n return { content: textContent, tool_calls: toolCalls.length > 0 ? toolCalls : undefined };\n}\n",[],[365,786],"d4e758a2-5d61-11f1-be46-1607ee4eb77e","knowledge_seed.ts","d4e75974-5d61-11f1-be46-1607ee4eb77e",579,["D",1780281019016],"// W2 — Knowledge seed runner (Part 4 + Part 5 corrections).\n// `math_knowledge_seed.ts` only DEFINES MATH_KNOWLEDGE_SEEDS — nothing ever inserted it,\n// and the growth-model pack (Part 4) had no home. This module is the missing wire: it\n// inserts both packs into knowledge_files on boot, idempotently, gated by a version key in\n// app_config so it runs once (and re-runs only when the pack version bumps).\nimport { sqlite } from \"../db.ts\";\nimport { MATH_KNOWLEDGE_SEEDS } from \"./math_knowledge_seed.ts\";\n\nconst SEED_VERSION = 2; // bump to force a re-seed of new/changed rows\n\ninterface SeedRow { name: string; kind: string; targets: string[]; content: string; source: string }\n\n// Part 4 — growth-model mental models. Dense rules + thresholds + decision heuristics, ending\n// in an \"App wiring\" line so the AI knows how the engine should use each.\nconst GROWTH_MODEL_SEEDS: Array<{ name: string; targets: string[]; content: string }> = [\n {\n name: \"Growth loops > funnels\",\n targets: [\"chat\", \"recommendations\", \"correlations\"],\n content: `### Growth loops beat funnels\n- Funnels grow LINEARLY (must be re-fed forever); loops COMPOUND (output re-feeds input). Always ask \"how does one cohort produce the next?\"\n- Common SaaS loops: Content (publish → rank → traffic → signups → UGC/links → rank), Referral/Viral (users invite users; k-factor = invites × conversion-rate), Paid (spend → customers → revenue → reinvest; compounds only if payback < cycle), Sales (win → case study/referral → win).\n- A loop is healthy when its coefficient ≥ 1 over a reasonable cycle; below 1 it decays without external feeding.\n- App wiring: when recommending, prefer actions that strengthen a loop's weakest step over one-off funnel bumps; annotate correlations with which loop they belong to.`,\n },\n {\n name: \"Retention is the foundation\",\n targets: [\"funnel_summary\", \"kpis\", \"recommendations\", \"cohort_ltv\", \"retention\"],\n content: `### Retention is the foundation (the silent killer)\n- Retention compounds and is slow to reveal (judge over ≥12 months). Poor retention silently caps every acquisition dollar.\n- Watch the retention CURVE SHAPE: it must FLATTEN (a stable plateau = product-market fit); a curve trending to zero = leaky bucket → fix before scaling acquisition.\n- Benchmarks: GRR ≥ 90% (good); NRR ≥ 110–120% world-class at scale; < 100% NRR at scale = structural stickiness problem.\n- Fix order: activation → engagement (habit/loop) → resurrection → monetization expansion, before pouring in top-of-funnel.\n- App wiring: make NRR/GRR a North Star input; if the retention curve isn't flattening, down-rank acquisition recommendations.`,\n },\n {\n name: \"North Star + 3–5 inputs\",\n targets: [\"kpis\", \"recommendations\", \"chat\"],\n content: `### North Star Metric + 3–5 inputs\n- NSM = a LEADING indicator of value delivered (not lagging revenue). It must be movable via 3–5 input metrics that teams influence daily and that ladder up.\n- Canonical input shape: Breadth (# active) × Depth (actions/user) × Frequency (sessions/period) × Efficiency (conversion / time-to-value).\n- Avoid vanity NSMs (registered users, pageviews) and pure-lagging NSMs (MRR).\n- App wiring: require the NSM to declare inputs; show each input's contribution + which is the current constraint; pace on_track by input trajectory, not a straight line.`,\n },\n {\n name: \"AARRR as a diagnostic lens\",\n targets: [\"funnel_summary\", \"correlations\"],\n content: `### AARRR as a diagnostic lens (paired with loops)\n- Use Acquisition → Activation → Retention → Referral → Revenue to LOCATE the constraint; then attach a LOOP to relieve it.\n- Diagnose in reverse: weak Revenue often traces to weak Activation/Retention upstream — fix upstream first.\n- App wiring: keep the AARRR tag on every metric, but report the BINDING CONSTRAINT stage, not an average \"health\".`,\n },\n {\n name: \"Signal-based / account-based GTM (MQL is dead)\",\n targets: [\"lead_scoring\", \"chat\", \"recommendations\", \"accounts\"],\n content: `### Signal-based / account-based GTM\n- MQL is a LAGGING indicator; score ACCOUNTS, not individuals (B2B buys in committees).\n- Lead/Account score = FIT × INTENT. Fit = ICP firmographics (industry, size, role). Intent = TIME-DECAYED weighted signals (pricing visit, demo/proposal ask, reply, repeat visits, multiple contacts from one domain).\n- Apply DECAY (intent rots in days/weeks) and DEDUPE to the account. A single meeting ≠ a hot account; a cluster of decaying signals from multiple contacts = hot.\n- PQL (product-qualified) converts 5–6× MQL → prioritize product-usage signals when available.\n- App wiring: roll lead scores up by domain and multiply by an ICP-fit score derived from Company Profile / Brand Context.`,\n },\n {\n name: \"Unit economics that matter (2025–26 benchmarks)\",\n targets: [\"kpis\", \"recommendations\", \"attribution\"],\n content: `### Unit economics that matter\n- LTV:CAC ≥ 3 viable; top quartile 4–6×. CAC payback < 12–18 mo (PLG/SMB < 12). NRR ≥ 110–120% at scale. Rule of 40 (growth% + FCF margin%) ≥ 40, target 50+. Magic Number (net-new ARR / prior-Q S&M) > 0.75 = efficient.\n- LTV must use GROSS MARGIN and the retention curve, not ARPU × arbitrary months. CAC = fully-loaded acquisition spend ÷ new customers (not goal completions).\n- App wiring: compute from the Revenue sheet + spend; grade against these thresholds with healthy/borderline/unhealthy; flag any figure that is an estimate.`,\n },\n {\n name: \"Bullseye channel strategy\",\n targets: [\"recommendations\", \"chat\", \"attribution\"],\n content: `### Bullseye channel strategy\n- One channel DOMINATES acquisition at a time. Process: Brainstorm all 19 channels → Rank (inner/potential/longshot) → Prioritize 3 → Test cheaply (cost-per-acq? volume available? right customers?) → Focus on the winner; revisit when it saturates.\n- Don't diversify channels prematurely; diversify LOOPS, not unproven channels.\n- App wiring: a channel-test ledger (cost/volume/fit per channel from attribution) → recommend doubling the current inner-ring channel, testing the next 3, dropping longshots.`,\n },\n {\n name: \"Experiment rigor\",\n targets: [\"ice_suggest\", \"recommendations\"],\n content: `### Experiment rigor\n- Every test needs: a hypothesis (because X, if we do Y, metric Z moves by W), ONE success metric, a baseline, an MDE (minimum detectable effect), a sample size, and a SIGNIFICANCE call (don't read results before powered).\n- A result is a WIN only if significant; otherwise inconclusive. Learnings compound — feed realized win-rates back as priors so ICE confidence reflects evidence, not vibes.\n- Prefer high-ICE × addresses-the-binding-constraint; kill fast, document always.\n- App wiring: completeExperiment computes significance + writes the metric back via propagate; suggestICE uses win-rate-by-lever as its confidence prior.`,\n },\n];\n\nfunction estTokens(s: string): number { return Math.ceil((s || \"\").length / 4); }\n\nasync function getSeedVersion(): Promise<number> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = 'knowledge_seed_version'\");\n const raw = (r.rows?.[0] as any[])?.[0];\n if (raw == null) return 0;\n return parseInt(String(raw).replace(/^\"|\"$/g, \"\"), 10) || 0;\n } catch { return 0; }\n}\n\nasync function rowExists(name: string, source: string): Promise<boolean> {\n try {\n const r = await sqlite.execute(\"SELECT 1 FROM knowledge_files WHERE name = ? AND source = ? LIMIT 1\", [name, source]);\n return (r.rows?.length || 0) > 0;\n } catch { return false; }\n}\n\nasync function insertSeed(row: SeedRow): Promise<void> {\n // Idempotent: skip if a row with the same name+source already exists.\n if (await rowExists(row.name, row.source)) return;\n await sqlite.execute(\n `INSERT INTO knowledge_files (name, kind, content, summary, targets, source, tokens_estimate, enabled)\n VALUES (?, ?, ?, ?, ?, ?, ?, 1)`,\n [row.name, row.kind, row.content, row.content.slice(0, 1400), JSON.stringify(row.targets), row.source, estTokens(row.content)]\n );\n}\n\n// Inserts both packs once (gated by knowledge_seed_version). Safe to call on every boot.\nexport async function seedKnowledge(): Promise<{ seeded: number; skipped: boolean }> {\n try {\n const current = await getSeedVersion();\n if (current >= SEED_VERSION) return { seeded: 0, skipped: true };\n\n let seeded = 0;\n for (const m of MATH_KNOWLEDGE_SEEDS) {\n await insertSeed({ name: m.name, kind: m.kind, targets: m.targets, content: m.content, source: \"seed_math\" });\n seeded++;\n }\n for (const g of GROWTH_MODEL_SEEDS) {\n await insertSeed({ name: g.name, kind: \"growth-model\", targets: g.targets, content: g.content, source: \"seed_growth\" });\n seeded++;\n }\n\n await sqlite.execute(\n `INSERT INTO app_config (key, value) VALUES ('knowledge_seed_version', ?)\n ON CONFLICT(key) DO UPDATE SET value = excluded.value`,\n [String(SEED_VERSION)]\n );\n return { seeded, skipped: false };\n } catch (e) {\n // Never let seeding break boot.\n console.log(\"[knowledge_seed] failed:\", String((e as any)?.message || e));\n return { seeded: 0, skipped: true };\n }\n}\n",[],[365,795],"28442654-5d69-11f1-8587-1607ee4eb77e","knowledge.ts","692223bc-5989-11f1-aff7-ee650bb23af1",619,["D",1780284165312],"// Knowledge files — growth playbooks/frameworks AND math formulas. Each file targets one\n// or more AI features (chat, kpis, ice_suggest, funnel_summary, etc.) — when a feature\n// runs, only its targeted knowledge is injected. Empty targets = global (affects every\n// chat turn). Storage: knowledge_files table.\n//\n// Designed for token-efficient injection: AI summaries capped to ~500 tokens each; consumer\n// concatenates targeted+enabled summaries up to a per-call budget.\nimport { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\n// Registry of AI features that can be targeted by a knowledge file. Used by the graph UI\n// to render feature nodes and by chat/intelligence to look up applicable knowledge.\nexport const AI_FEATURES: Array<{ id: string; label: string; description: string }> = [\n { id: \"chat\", label: \"Chat Assistant\", description: \"General chat. Reads global + chat-targeted knowledge on every turn.\" },\n { id: \"funnel_summary\", label: \"Funnel Health\", description: \"TOFU/MOFU/BOFU stage scoring + overall composite.\" },\n { id: \"kpis\", label: \"AIO KPIs\", description: \"CAC, LTV, MER, Pipeline Coverage, funnel conversion.\" },\n { id: \"ice_suggest\", label: \"ICE Suggester\", description: \"LLM-assisted ICE/RICE/WSJF scoring for experiments.\" },\n { id: \"lead_scoring\", label: \"Lead Scoring\", description: \"MQL/SQL classification + intent signal weights.\" },\n { id: \"digest\", label: \"Weekly Digest\", description: \"Growth digest synthesis.\" },\n { id: \"standup\", label: \"Daily Standup\", description: \"Daily team brief.\" },\n { id: \"cohort_ltv\", label: \"Cohort LTV\", description: \"Cohort survival LTV computation.\" },\n { id: \"correlations\", label: \"Signal Correlations\", description: \"Cross-metric causal heuristics.\" },\n { id: \"recommendations\", label: \"Recommendations\", description: \"Top-3 next-best-action ranking.\" },\n];\nconst FEATURE_IDS = new Set(AI_FEATURES.map(f => f.id));\n\nexport interface KnowledgeFile {\n id: number;\n name: string;\n kind: string;\n content: string;\n summary: string | null;\n targets: string[]; // empty = global (affects every feature)\n source: string; // 'user' | 'seed_math'\n tokens_estimate: number;\n enabled: boolean;\n uploaded_at: string;\n updated_at: string;\n}\n\nfunction parseTargets(raw: any): string[] {\n if (Array.isArray(raw)) return raw.filter(t => typeof t === \"string\");\n if (typeof raw === \"string\" && raw.trim()) {\n try { const parsed = JSON.parse(raw); return Array.isArray(parsed) ? parsed.filter(t => typeof t === \"string\") : []; } catch { return []; }\n }\n return [];\n}\nfunction sanitizeTargets(targets: any): string[] {\n // Drop invalid feature ids (registry is the source of truth).\n return parseTargets(targets).filter(t => FEATURE_IDS.has(t));\n}\n\n// Rough chars/4 token estimator — good enough for prompt-budget caps without a tokenizer dep.\nconst estTokens = (s: string): number => Math.ceil((s || \"\").length / 4);\n\nfunction rowToFile(cols: string[], row: any[]): KnowledgeFile {\n const obj: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) obj[cols[i]] = row[i];\n return {\n id: Number(obj.id),\n name: String(obj.name),\n kind: String(obj.kind || \"playbook\"),\n content: String(obj.content || \"\"),\n summary: obj.summary ? String(obj.summary) : null,\n targets: parseTargets(obj.targets),\n source: String(obj.source || \"user\"),\n tokens_estimate: Number(obj.tokens_estimate || 0),\n enabled: Number(obj.enabled || 0) === 1,\n uploaded_at: String(obj.uploaded_at || \"\"),\n updated_at: String(obj.updated_at || \"\"),\n };\n}\n\n// ─── Listing — light (no content body) for the Sidebar / file list ──────────\nexport async function listKnowledgeFiles(): Promise<Array<Omit<KnowledgeFile, \"content\">>> {\n try {\n const r = await sqlite.execute(\n `SELECT id, name, kind, summary, targets, source, tokens_estimate, enabled, uploaded_at, updated_at\n FROM knowledge_files\n ORDER BY source = 'seed_math' DESC, uploaded_at DESC`\n );\n return (r.rows || []).map((row: any) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < r.columns.length; i++) obj[r.columns[i]] = row[i];\n return {\n id: Number(obj.id),\n name: String(obj.name),\n kind: String(obj.kind || \"playbook\"),\n summary: obj.summary ? String(obj.summary) : null,\n targets: parseTargets(obj.targets),\n source: String(obj.source || \"user\"),\n tokens_estimate: Number(obj.tokens_estimate || 0),\n enabled: Number(obj.enabled || 0) === 1,\n uploaded_at: String(obj.uploaded_at || \"\"),\n updated_at: String(obj.updated_at || \"\"),\n };\n });\n } catch {\n return [];\n }\n}\n\nexport async function getKnowledgeFile(id: string | number): Promise<KnowledgeFile | null> {\n try {\n const r = await sqlite.execute(\n `SELECT id, name, kind, content, summary, targets, source, tokens_estimate, enabled, uploaded_at, updated_at\n FROM knowledge_files WHERE id = ?`,\n [Number(id)]\n );\n if (!r.rows?.length) return null;\n return rowToFile(r.columns, r.rows[0] as any[]);\n } catch {\n return null;\n }\n}\n\n// ─── Create — accepts file content (already decoded). Generates AI summary opportunistically. ───\nexport async function createKnowledgeFile(data: {\n name: string;\n kind?: string;\n content: string;\n enabled?: boolean;\n targets?: string[];\n}): Promise<KnowledgeFile | null> {\n const name = (data.name || \"Untitled\").trim();\n const kind = (data.kind || \"playbook\").trim();\n const content = String(data.content || \"\");\n if (!content.trim()) return null;\n const tokens = estTokens(content);\n const enabled = data.enabled === false ? 0 : 1;\n const targets = JSON.stringify(sanitizeTargets(data.targets));\n\n await sqlite.execute(\n `INSERT INTO knowledge_files (name, kind, content, tokens_estimate, enabled, targets, source)\n VALUES (?, ?, ?, ?, ?, ?, 'user')`,\n [name, kind, content, tokens, enabled, targets]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = Number((idRow.rows[0] as any[])[0]);\n\n // Best-effort AI summary so future chats inject the gist (not the raw file).\n // We don't block the create call — fire-and-forget. If LLM fails, summary stays null.\n summarizeKnowledgeFile(id).catch(() => {});\n\n return await getKnowledgeFile(id);\n}\n\nexport async function updateKnowledgeFile(\n id: string | number,\n data: { name?: string; kind?: string; enabled?: boolean; content?: string; targets?: string[] }\n): Promise<KnowledgeFile | null> {\n const cur = await getKnowledgeFile(id);\n if (!cur) return null;\n const name = data.name?.trim() || cur.name;\n const kind = data.kind?.trim() || cur.kind;\n const enabled = typeof data.enabled === \"boolean\" ? (data.enabled ? 1 : 0) : (cur.enabled ? 1 : 0);\n const content = typeof data.content === \"string\" ? data.content : cur.content;\n const tokens = estTokens(content);\n const targets = JSON.stringify(\n Array.isArray(data.targets) ? sanitizeTargets(data.targets) : cur.targets\n );\n await sqlite.execute(\n `UPDATE knowledge_files\n SET name = ?, kind = ?, content = ?, tokens_estimate = ?, enabled = ?, targets = ?, updated_at = datetime('now')\n WHERE id = ?`,\n [name, kind, content, tokens, enabled, targets, Number(id)]\n );\n // Re-summarize when content changed.\n if (typeof data.content === \"string\" && data.content !== cur.content) {\n summarizeKnowledgeFile(id).catch(() => {});\n }\n return await getKnowledgeFile(id);\n}\n\nexport async function deleteKnowledgeFile(id: string | number): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM knowledge_files WHERE id = ?\", [Number(id)]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── Summarization — LLM compresses content into a 300-500 token gist for chat injection ─\nexport async function summarizeKnowledgeFile(id: string | number): Promise<string | null> {\n const file = await getKnowledgeFile(id);\n if (!file || !file.content?.trim()) return null;\n\n const cfg = await getLLMConfig();\n if (!cfg.api_key) return null;\n\n // Short-circuit: if content is already <2000 chars, the content IS the summary.\n if (file.content.length < 2000) {\n const summary = file.content.trim();\n await sqlite.execute(\n \"UPDATE knowledge_files SET summary = ?, updated_at = datetime('now') WHERE id = ?\",\n [summary, Number(id)]\n );\n return summary;\n }\n\n const prompt = `You compress growth playbooks for an AI assistant. The assistant will read your\nsummary on every chat turn, so be dense and information-rich. Preserve:\n- Frameworks, principles, scoring rules, formulas\n- Concrete heuristics, numbers, thresholds\n- Decision rules (\"when X → do Y\")\n- Anti-patterns to avoid\nDrop: storytelling, examples that don't carry a rule, marketing fluff.\n\nOutput: 400-600 words. Markdown bullets. No preamble.\n\n---\nTITLE: ${file.name}\nKIND: ${file.kind}\n---\n${file.content.slice(0, 50000)}`;\n\n try {\n const resp = await callLLM(\n cfg,\n [{ role: \"user\", content: prompt }],\n { maxTokens: 1200 }\n );\n const summary = (resp.content || \"\").trim();\n if (!summary) return null;\n await sqlite.execute(\n \"UPDATE knowledge_files SET summary = ?, updated_at = datetime('now') WHERE id = ?\",\n [summary, Number(id)]\n );\n return summary;\n } catch {\n return null;\n }\n}\n\n// ─── Targeted injection helper — only loads knowledge for THIS feature ────────\n// Each feature can pull ONLY the knowledge files that listed it in `targets`, plus any\n// globally-targeted files (empty targets array). This keeps prompts lean per feature.\n//\n// Usage:\n// const ctx = await loadEnabledKnowledgeContext({ target: 'kpis', maxChars: 4000 });\n// // → returns markdown with only KPI-relevant playbooks\n//\n// Back-compat: calling with no target (or 'chat') returns global + chat-targeted files,\n// matching the previous behavior api/chat.ts depends on.\nexport async function loadEnabledKnowledgeContext(\n opts?: { target?: string; maxChars?: number } | number\n): Promise<string> {\n // Back-compat: prior signature was loadEnabledKnowledgeContext(maxChars)\n const maxChars = typeof opts === \"number\" ? opts : (opts?.maxChars ?? 6000);\n const target = typeof opts === \"object\" ? opts?.target : undefined;\n\n try {\n const r = await sqlite.execute(\n `SELECT name, kind, summary, content, targets\n FROM knowledge_files\n WHERE enabled = 1\n ORDER BY uploaded_at DESC`\n );\n const rows = (r.rows || []) as any[];\n if (rows.length === 0) return \"\";\n\n // Filter by target: include the file if its targets are EMPTY (global) or contain the requested target.\n const filtered = target\n ? rows.filter((row) => {\n const t = parseTargets(row[4]);\n return t.length === 0 || t.includes(target);\n })\n : rows;\n if (filtered.length === 0) return \"\";\n\n const header = target\n ? `## Active Growth Knowledge (filtered for \"${target}\")\\nApply the rules and formulas below when reasoning.\\n\\n`\n : \"## Active Growth Knowledge\\nThe user has uploaded these playbooks. Apply them when reasoning.\\n\\n\";\n let out = header;\n let remaining = maxChars - out.length;\n for (const row of filtered) {\n const name = String(row[0]);\n const kind = String(row[1] || \"playbook\");\n const summary = row[2] ? String(row[2]) : null;\n const content = row[3] ? String(row[3]) : \"\";\n const body = summary || content.slice(0, 800);\n const block = `### ${name} (${kind})\\n${body}\\n\\n`;\n if (block.length > remaining) {\n if (remaining > 200) out += block.slice(0, remaining - 12) + \"\\n…[trunc]\\n\";\n break;\n }\n out += block;\n remaining -= block.length;\n }\n return out.trim();\n } catch {\n return \"\";\n }\n}\n\n// ─── Knowledge graph builder — nodes + edges for the visual UI ───────────────\n// Used by GET /api/knowledge/graph. Returns:\n// - features: the AI_FEATURES registry (consumer reference)\n// - nodes: knowledge_files as 'knowledge' nodes + AI_FEATURES as 'feature' nodes\n// - edges: one edge per (file, target) pair. GLOBAL files (empty targets) get an\n// edge to every feature so the UI can render them as connected-to-all.\nexport async function buildKnowledgeGraph(): Promise<{\n features: typeof AI_FEATURES;\n nodes: Array<any>;\n edges: Array<{ from: number; to: string; global?: boolean }>;\n}> {\n const files = await listKnowledgeFiles();\n const featureNodes = AI_FEATURES.map((f) => ({\n id: \"feature:\" + f.id,\n type: \"feature\" as const,\n label: f.label,\n description: f.description,\n feature_id: f.id,\n }));\n const knowledgeNodes = files.map((f) => ({\n id: \"knowledge:\" + f.id,\n type: \"knowledge\" as const,\n label: f.name,\n knowledge_id: f.id,\n kind: f.kind,\n source: (f as any).source,\n enabled: f.enabled,\n targets: f.targets,\n tokens: f.tokens_estimate,\n }));\n const edges: Array<{ from: number; to: string; global?: boolean }> = [];\n for (const f of files) {\n if (!f.enabled) continue;\n if (f.targets.length === 0) {\n // GLOBAL — connect to every feature\n for (const feat of AI_FEATURES) {\n edges.push({ from: f.id, to: feat.id, global: true });\n }\n } else {\n for (const t of f.targets) {\n edges.push({ from: f.id, to: t });\n }\n }\n }\n return {\n features: AI_FEATURES,\n nodes: [...knowledgeNodes, ...featureNodes],\n edges,\n };\n}\n",[],[365,803],"27c0a900-5d69-11f1-ac6f-1607ee4eb77e","intelligence.ts","d5cf99da-580e-11f1-b0e3-ee650bb23af1",618,["D",1780284164483],12,"// Cross-Metric Correlation Engine + Unified Funnel + LLM-generated commentary feed.\n// Maps all 14 integrations to TOFU/MOFU/BOFU stages, fetches live metrics,\n// computes health scores against 30d baselines, surfaces causal correlations\n// with narrative + action prompts. Cached aggressively (5min TTL) since fetches\n// hit external APIs.\nimport { sqlite } from \"../db.ts\";\nimport { getCredentials, withGoogleFallback, safeJson } from \"./integrations/_shared.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\n// ─── FUNNEL MAPPING ─────────────────────────────────────────────────────────\ntype Stage = \"TOFU\" | \"MOFU\" | \"BOFU\";\n\nexport type AARRR = \"acquisition\" | \"activation\" | \"retention\" | \"referral\" | \"revenue\";\n\nexport interface MetricPoint {\n integration: string;\n metric: string;\n value: number;\n unit: string;\n stage: Stage;\n aarrr?: AARRR; // Pirate Metrics category\n recorded_at?: string; // ISO timestamp from metric_history\n}\n\n// Absolute health benchmarks per metric: maps raw value → 0-100 health.\n// Replaces the warm-start \"always 50%\" bug — gives real per-metric scoring based\n// on industry SaaS benchmarks. Returns null if no benchmark defined (falls back to\n// baseline-relative scoring).\nfunction normalizeAbsolute(integration: string, metric: string, value: number): number | null {\n const k = `${integration}:${metric}`;\n const clamp = (n: number) => Math.max(0, Math.min(100, n));\n switch (k) {\n // SEO authority (Ahrefs) — DR is already 0-100; refdomains log-scaled\n case \"ahrefs:domain_rating\": return clamp(value);\n case \"ahrefs:referring_domains\": return clamp(Math.log10(value + 1) * 25);\n // Search Console — CTR 0-5% maps to 0-100, position 1 = 100 falling 5%/rank\n case \"gsc:ctr\": return clamp((value / 5) * 100);\n case \"gsc:impressions\": return clamp(Math.log10(value + 1) * 20);\n case \"gsc:clicks\": return clamp(Math.log10(value + 1) * 25);\n case \"gsc:average_position\": return clamp(100 - (value - 1) * 5);\n // GA4 — engagement_time/180s = 100, conversions sqrt curve, bounce inverted\n case \"ga4:sessions\": return clamp(Math.log10(value + 1) * 20);\n case \"ga4:engagement_time\": return clamp((value / 180) * 100);\n case \"ga4:pages_per_session\": return clamp(((value - 1) / 3) * 100);\n case \"ga4:bounce_rate\": return clamp(100 - value);\n case \"ga4:conversions\": return clamp(Math.sqrt(value) * 15);\n // YouTube\n case \"youtube:views\": return clamp(Math.log10(value + 1) * 15);\n case \"youtube:subscribers\": return clamp(Math.log10(value + 1) * 20);\n // WordPress content velocity — 20 posts/period = full health\n case \"wordpress:published_posts\": return clamp(value * 5);\n // Communications volume — log-scaled\n case \"gmail:messages_total\": return clamp(Math.log10(value + 1) * 15);\n case \"gmail:threads_total\": return clamp(Math.log10(value + 1) * 18);\n case \"gmail:deals_initiated\": return clamp(value * 10);\n case \"outlook:inbox_count\": return clamp(Math.log10(value + 1) * 15);\n case \"outlook:meetings_booked\":return clamp(value * 10);\n // LinkedIn\n case \"linkedin:post_impressions\": return clamp(Math.log10(value + 1) * 15);\n default: return null; // no absolute benchmark → fall back to baseline-relative\n }\n}\n\n// AARRR mapping — same metric can belong to multiple categories.\n// gsc.average_position, gsc.ctr, youtube.subscribers, linkedin.reply_rate all carry\n// Pirate-Metric classification so the funnel summary never leaves stages undefined.\nconst AARRR_MAP: Record<string, AARRR> = {\n // Acquisition (first touch)\n \"gsc:clicks\": \"acquisition\", \"youtube:views\": \"acquisition\", \"linkedin:post_impressions\": \"acquisition\",\n \"gsc:impressions\": \"acquisition\", \"ahrefs:organic_keywords\": \"acquisition\",\n \"gsc:average_position\": \"acquisition\",\n // Activation (first meaningful engagement)\n \"ga4:sessions\": \"activation\", \"wordpress:published_posts\": \"activation\",\n \"gmail:messages_total\": \"activation\", \"gsc:ctr\": \"activation\",\n \"linkedin:reply_rate\": \"activation\",\n // Retention (repeated engagement)\n \"ga4:engagement_time\": \"retention\", \"ga4:pages_per_session\": \"retention\", \"ga4:bounce_rate\": \"retention\",\n \"gmail:threads_total\": \"retention\", \"youtube:subscribers\": \"retention\",\n // Referral (audience amplification)\n \"ahrefs:referring_domains\": \"referral\", \"ahrefs:domain_rating\": \"referral\",\n // Revenue (monetization events)\n \"ga4:conversions\": \"revenue\", \"outlook:inbox_count\": \"revenue\",\n \"outlook:meetings_booked\": \"revenue\", \"gmail:deals_initiated\": \"revenue\",\n};\nconst aarrrOf = (integration: string, metric: string): AARRR | undefined =>\n AARRR_MAP[`${integration}:${metric}`];\n\n// ─── BAYESIAN-INSPIRED CONFIDENCE ────────────────────────────────────────────\n// Penalize stale data, missing integrations, and rules currently inside their lag window.\nexport function computeConfidence(args: {\n source_freshness_days: number;\n target_freshness_days: number;\n source_has_data: boolean;\n target_has_data: boolean;\n lag_days: number;\n current_day_offset: number;\n}): number {\n let c = 1.0;\n if (args.source_freshness_days > 7) c *= 0.8;\n if (args.source_freshness_days > 30) c *= 0.6;\n if (args.target_freshness_days > 7) c *= 0.8;\n if (!args.source_has_data) c *= 0.3;\n if (!args.target_has_data) c *= 0.3;\n if (args.current_day_offset < args.lag_days) c *= 0.5;\n return Math.round(c * 100) / 100;\n}\n\n// ─── METRIC HISTORY (timeseries for delta_7d) ────────────────────────────────\nasync function recordMetric(integration: string, metric: string, value: number): Promise<void> {\n try {\n await sqlite.execute(\n \"INSERT INTO metric_history (integration, metric_key, value) VALUES (?, ?, ?)\",\n [integration, metric, value]\n );\n // Prune anything older than 60d to keep table bounded.\n await sqlite.execute(\n \"DELETE FROM metric_history WHERE recorded_at < datetime('now','-60 days')\"\n );\n } catch { /* non-critical */ }\n}\n\nasync function getMetricValueDaysAgo(integration: string, metric: string, days: number): Promise<number | null> {\n try {\n // Find closest recorded value to N days ago (within ±2 days window).\n const r = await sqlite.execute(\n `SELECT value FROM metric_history\n WHERE integration = ? AND metric_key = ?\n AND recorded_at <= datetime('now', ?)\n ORDER BY recorded_at DESC LIMIT 1`,\n [integration, metric, `-${days} days`]\n );\n if (!r.rows?.length) return null;\n return Number((r.rows[0] as any[])[0]);\n } catch { return null; }\n}\n\nexport interface CorrelationSignal {\n id: string;\n source_metric: { integration: string; metric: string; value: number; unit: string };\n target_metric: { integration: string; metric: string; value: number; unit: string };\n direction: \"positive\" | \"negative\" | \"neutral\";\n strength: \"strong\" | \"moderate\" | \"weak\";\n lag_days: number;\n narrative: string;\n funnel_stage_from: Stage;\n funnel_stage_to: Stage;\n confidence: number;\n action_prompt: string;\n}\n\nexport interface StageHealth {\n stage: Stage;\n label: string;\n description: string;\n metrics: Array<MetricPoint & { delta_pct: number | null; baseline: number | null; delta_7d?: number | null; delta_30d?: number | null }>;\n health_score: number; // 0-100\n diagnosis: string;\n}\n\nexport interface FunnelSummary {\n stages: StageHealth[];\n bottleneck: { stage: Stage; reason: string } | null;\n // Weighted composite — TOFU 0.25 · MOFU 0.40 · BOFU 0.35. BOFU revenue\n // matters more than top-of-funnel volume, MOFU drives qualified pipeline.\n overall_health: number;\n total_metrics_observed: number;\n total_integrations_connected: number;\n generated_at: string;\n}\n\nexport interface IntelligenceCard {\n type: \"correlation\" | \"anomaly\" | \"opportunity\" | \"risk\" | \"win\";\n priority: \"critical\" | \"high\" | \"medium\";\n title: string;\n body: string;\n metrics: string[];\n funnel_stages: Stage[];\n agent_name: string;\n timestamp: string;\n action_label: string;\n action_payload: { question: string; context?: any };\n}\n\n// ─── CACHE HELPERS ──────────────────────────────────────────────────────────\nasync function getCached(key: string): Promise<any | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, expires_at FROM intelligence_cache WHERE cache_key = ?\",\n [key]\n );\n if (!r.rows?.length) return null;\n const [payload, expiresAt] = r.rows[0] as any[];\n if (new Date(String(expiresAt)).getTime() < Date.now()) return null;\n return JSON.parse(String(payload));\n } catch { return null; }\n}\n\n// Read generated_at so the feed can invalidate itself when upstream summary is fresher.\nasync function getCachedTimestamp(key: string): Promise<{ payload: any; generated_at: number; expires_at: number } | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, generated_at, expires_at FROM intelligence_cache WHERE cache_key = ?\",\n [key]\n );\n if (!r.rows?.length) return null;\n const [payload, generatedAt, expiresAt] = r.rows[0] as any[];\n const exp = new Date(String(expiresAt)).getTime();\n if (exp < Date.now()) return null;\n return {\n payload: JSON.parse(String(payload)),\n generated_at: new Date(String(generatedAt)).getTime(),\n expires_at: exp,\n };\n } catch { return null; }\n}\n\nasync function setCached(key: string, payload: any, ttlSeconds: number): Promise<void> {\n const expiresAt = new Date(Date.now() + ttlSeconds * 1000).toISOString();\n await sqlite.execute(\n \"INSERT OR REPLACE INTO intelligence_cache (cache_key, payload, generated_at, expires_at) VALUES (?, ?, datetime('now'), ?)\",\n [key, JSON.stringify(payload), expiresAt]\n );\n}\n\nasync function getBaseline(integration: string, metricKey: string): Promise<number | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT baseline_value FROM metric_baselines WHERE integration = ? AND metric_key = ?\",\n [integration, metricKey]\n );\n if (!r.rows?.length) return null;\n return Number((r.rows[0] as any[])[0]);\n } catch { return null; }\n}\n\nasync function setBaseline(integration: string, metricKey: string, value: number): Promise<void> {\n await sqlite.execute(\n \"INSERT OR REPLACE INTO metric_baselines (integration, metric_key, baseline_value, last_updated) VALUES (?, ?, ?, datetime('now'))\",\n [integration, metricKey, value]\n );\n}\n\n// Rolling 30-day baseline. Replaces the frozen warm-start baseline once we have\n// at least 3 samples in metric_history, so health scoring drifts with reality instead\n// of forever comparing against the first observed value.\nasync function getRollingBaseline(integration: string, metricKey: string): Promise<{ avg: number; sample_count: number } | null> {\n try {\n const r = await sqlite.execute(\n `SELECT AVG(value) AS avg_v, COUNT(*) AS n\n FROM metric_history\n WHERE integration = ? AND metric_key = ?\n AND recorded_at >= datetime('now','-30 days')`,\n [integration, metricKey]\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any;\n const avg = Array.isArray(row) ? Number(row[0]) : Number(row.avg_v);\n const n = Array.isArray(row) ? Number(row[1]) : Number(row.n);\n if (!Number.isFinite(avg) || !Number.isFinite(n) || n < 3) return null;\n return { avg, sample_count: n };\n } catch { return null; }\n}\n\n// Persist the rolling figures back to metric_baselines so external readers see them too.\nasync function persistRolling(integration: string, metricKey: string, rolling: { avg: number; sample_count: number }): Promise<void> {\n try {\n await sqlite.execute(\n \"UPDATE metric_baselines SET rolling_avg = ?, sample_count = ?, last_updated = datetime('now') WHERE integration = ? AND metric_key = ?\",\n [rolling.avg, rolling.sample_count, integration, metricKey]\n );\n } catch { /* non-critical */ }\n}\n\n// ─── METRIC FETCHERS (live data from integrations) ──────────────────────────\n// Each returns MetricPoint[] or [] on error/missing creds. Never throws.\n\nasync function fetchAhrefs(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"ahrefs\");\n if (!creds?.api_key || !creds.extra_config?.target) return [];\n const target = creds.extra_config.target;\n const today = new Date().toISOString().slice(0, 10);\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(`https://api.ahrefs.com/v3/site-explorer/domain-rating?target=${encodeURIComponent(target)}&date=${today}`,\n { headers: { Authorization: `Bearer ${creds.api_key}` } });\n if (r.ok) {\n const d = await safeJson(r);\n const dr = d?.domain_rating?.domain_rating;\n if (typeof dr === \"number\") out.push({ integration: \"ahrefs\", metric: \"domain_rating\", value: dr, unit: \"DR\", stage: \"TOFU\" });\n }\n } catch {}\n try {\n const r = await fetch(`https://api.ahrefs.com/v3/site-explorer/backlinks-stats?target=${encodeURIComponent(target)}&date=${today}`,\n { headers: { Authorization: `Bearer ${creds.api_key}` } });\n if (r.ok) {\n const d = await safeJson(r);\n const rd = d?.metrics?.refdomains;\n if (typeof rd === \"number\") out.push({ integration: \"ahrefs\", metric: \"referring_domains\", value: rd, unit: \"domains\", stage: \"TOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchGSC(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"gsc\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token || !creds.extra_config?.site_url) return [];\n const siteUrl = encodeURIComponent(creds.extra_config.site_url);\n const end = new Date().toISOString().slice(0, 10);\n const start = new Date(Date.now() - days * 864e5).toISOString().slice(0, 10);\n try {\n const r = await fetch(`https://searchconsole.googleapis.com/webmasters/v3/sites/${siteUrl}/searchAnalytics/query`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({ startDate: start, endDate: end, dimensions: [], rowLimit: 1 }),\n });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const row = d?.rows?.[0];\n if (!row) return [];\n return [\n { integration: \"gsc\", metric: \"impressions\", value: row.impressions || 0, unit: \"impressions\", stage: \"TOFU\" },\n { integration: \"gsc\", metric: \"clicks\", value: row.clicks || 0, unit: \"clicks\", stage: \"TOFU\" },\n { integration: \"gsc\", metric: \"ctr\", value: (row.ctr || 0) * 100, unit: \"%\", stage: \"MOFU\" },\n { integration: \"gsc\", metric: \"average_position\", value: row.position || 0, unit: \"pos\", stage: \"TOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchGA4(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"ga4\");\n creds = await withGoogleFallback(creds ?? {});\n const propertyId = creds.extra_config?.property_id;\n if (!creds.access_token || !propertyId) return [];\n try {\n const r = await fetch(`https://analyticsdata.googleapis.com/v1beta/properties/${propertyId}:runReport`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({\n metrics: [\n { name: \"sessions\" }, { name: \"userEngagementDuration\" }, { name: \"bounceRate\" },\n { name: \"conversions\" }, { name: \"screenPageViewsPerSession\" },\n ],\n dateRanges: [{ startDate: `${days}daysAgo`, endDate: \"today\" }],\n }),\n });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const row = d?.rows?.[0];\n if (!row) return [];\n const vals = row.metricValues || [];\n const num = (i: number) => parseFloat(vals[i]?.value || \"0\");\n const sessions = num(0);\n return [\n { integration: \"ga4\", metric: \"sessions\", value: sessions, unit: \"sessions\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"engagement_time\", value: sessions ? num(1) / sessions : 0, unit: \"sec/sess\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"bounce_rate\", value: num(2) * 100, unit: \"%\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"conversions\", value: num(3), unit: \"conv\", stage: \"BOFU\" },\n { integration: \"ga4\", metric: \"pages_per_session\", value: num(4), unit: \"pages\", stage: \"MOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchYouTube(): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"youtube\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token && !creds.api_key) return [];\n const auth = creds.access_token ? { Authorization: `Bearer ${creds.access_token}` } : {};\n const keyParam = !creds.access_token && creds.api_key ? `&key=${encodeURIComponent(creds.api_key)}` : \"\";\n try {\n const url = creds.access_token\n ? `https://www.googleapis.com/youtube/v3/channels?part=statistics&mine=true`\n : `https://www.googleapis.com/youtube/v3/channels?part=statistics&chart=mostPopular&maxResults=1${keyParam}`;\n const r = await fetch(url, { headers: auth });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const s = d?.items?.[0]?.statistics;\n if (!s) return [];\n return [\n { integration: \"youtube\", metric: \"views\", value: parseInt(s.viewCount || \"0\", 10), unit: \"views\", stage: \"TOFU\" },\n { integration: \"youtube\", metric: \"subscribers\", value: parseInt(s.subscriberCount || \"0\", 10), unit: \"subs\", stage: \"TOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchLinkedIn(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"linkedin\");\n if (!creds?.access_token) return [];\n try {\n const r = await fetch(\"https://api.linkedin.com/v2/me\", { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!r.ok) return [];\n return [\n { integration: \"linkedin\", metric: \"post_impressions\", value: 0, unit: \"impressions\", stage: \"TOFU\" },\n { integration: \"linkedin\", metric: \"reply_rate\", value: 0, unit: \"%\", stage: \"BOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchGmail(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"gmail\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token) return [];\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(\"https://gmail.googleapis.com/gmail/v1/users/me/profile\", {\n headers: { Authorization: `Bearer ${creds.access_token}` },\n });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"gmail\", metric: \"messages_total\", value: d?.messagesTotal || 0, unit: \"msgs\", stage: \"MOFU\" });\n out.push({ integration: \"gmail\", metric: \"threads_total\", value: d?.threadsTotal || 0, unit: \"threads\", stage: \"MOFU\" });\n }\n } catch {}\n // BOFU: deals_initiated — count Gmail threads with deal-intent keywords in the selected window.\n try {\n const q = encodeURIComponent(`newer_than:${days}d (pricing OR proposal OR contract OR quote OR \"sales agreement\")`);\n const r = await fetch(`https://gmail.googleapis.com/gmail/v1/users/me/messages?q=${q}&maxResults=1`,\n { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"gmail\", metric: \"deals_initiated\", value: d?.resultSizeEstimate || 0, unit: \"threads\", stage: \"BOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchOutlook(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"outlook\");\n if (!creds?.access_token) return [];\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(\"https://graph.microsoft.com/v1.0/me/messages?$top=1&$count=true\", {\n headers: { Authorization: `Bearer ${creds.access_token}`, ConsistencyLevel: \"eventual\" },\n });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"outlook\", metric: \"inbox_count\", value: d?.[\"@odata.count\"] || 0, unit: \"msgs\", stage: \"MOFU\" });\n }\n } catch {}\n // BOFU: meetings booked (next 14 days) — real calendar query\n try {\n const start = new Date().toISOString();\n const end = new Date(Date.now() + 14 * 86400000).toISOString();\n const r = await fetch(`https://graph.microsoft.com/v1.0/me/calendar/events?$top=1&$count=true&$filter=start/dateTime ge '${start}' and start/dateTime le '${end}'`,\n { headers: { Authorization: `Bearer ${creds.access_token}`, ConsistencyLevel: \"eventual\" } });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"outlook\", metric: \"meetings_booked\", value: d?.[\"@odata.count\"] || 0, unit: \"mtgs\", stage: \"BOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchWordPress(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"wordpress\");\n const siteUrl = (creds?.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl || !creds?.api_key) return [];\n try {\n const auth = btoa(`${creds.api_key}:${creds.api_secret ?? \"\"}`);\n const r = await fetch(`${siteUrl}/wp-json/wp/v2/posts?per_page=1`, {\n headers: { Authorization: `Basic ${auth}` },\n });\n if (!r.ok) return [];\n const total = parseInt(r.headers.get(\"x-wp-total\") || \"0\", 10);\n return [{ integration: \"wordpress\", metric: \"published_posts\", value: total, unit: \"posts\", stage: \"MOFU\" }];\n } catch { return []; }\n}\n\n// ─── FUNNEL SUMMARY ─────────────────────────────────────────────────────────\nconst STAGE_META: Record<Stage, { label: string; description: string }> = {\n TOFU: { label: \"Awareness & Reach\", description: \"Top-of-funnel visibility — SEO, social, content distribution.\" },\n MOFU: { label: \"Engagement & Consideration\", description: \"Mid-funnel signal — site engagement, content depth, outreach reply.\" },\n BOFU: { label: \"Conversion & Revenue\", description: \"Bottom-of-funnel — goal completions, deals, meetings, revenue.\" },\n};\n\nexport async function getFunnelSummary(force = false, days = 30): Promise<FunnelSummary> {\n const cacheKey = `funnel_summary_${days}d`;\n if (!force) {\n const cached = await getCached(cacheKey);\n if (cached) return cached;\n }\n\n const results = await Promise.allSettled([\n fetchAhrefs(), fetchGSC(days), fetchGA4(days), fetchYouTube(),\n fetchLinkedIn(), fetchGmail(days), fetchOutlook(), fetchWordPress(),\n ]);\n const allMetrics: MetricPoint[] = results\n .filter((r): r is PromiseFulfilledResult<MetricPoint[]> => r.status === \"fulfilled\")\n .flatMap((r) => r.value)\n .map((m) => ({ ...m, aarrr: aarrrOf(m.integration, m.metric), recorded_at: new Date().toISOString() }));\n\n // Update baselines on first run + write every value to metric_history for delta_7d\n for (const m of allMetrics) {\n const b = await getBaseline(m.integration, m.metric);\n if (b === null) await setBaseline(m.integration, m.metric, m.value);\n await recordMetric(m.integration, m.metric, m.value);\n }\n\n // Compute per-stage health. Each metric gets baseline delta + 7-day delta from metric_history.\n const stages: StageHealth[] = ([\"TOFU\", \"MOFU\", \"BOFU\"] as Stage[]).map((stage) => {\n const stageMetrics = allMetrics.filter((m) => m.stage === stage);\n const withDeltas = stageMetrics.map(async (m) => {\n // Prefer rolling 30d baseline once we have ≥3 samples; fall back to frozen warm-start.\n const frozen = await getBaseline(m.integration, m.metric);\n const rolling = await getRollingBaseline(m.integration, m.metric);\n const baseline = rolling ? rolling.avg : frozen;\n if (rolling) await persistRolling(m.integration, m.metric, rolling);\n const delta_pct = baseline && baseline !== 0 ? ((m.value - baseline) / Math.abs(baseline)) * 100 : null;\n // delta_7d / delta_30d: real timeseries comparison vs same metric N days ago (A2).\n // TOFU stage UI prefers 30d (SEO/branding moves slowly); MOFU/BOFU keep 7d for tighter feedback.\n const [v7, v30] = await Promise.all([\n getMetricValueDaysAgo(m.integration, m.metric, 7),\n getMetricValueDaysAgo(m.integration, m.metric, 30),\n ]);\n const delta_7d = v7 != null && v7 !== 0 ? ((m.value - v7) / Math.abs(v7)) * 100 : null;\n const delta_30d = v30 != null && v30 !== 0 ? ((m.value - v30) / Math.abs(v30)) * 100 : null;\n return { ...m, delta_pct, baseline, delta_7d, delta_30d };\n });\n return { stage, metrics: withDeltas, stageMetrics };\n }) as any;\n\n // Resolve the promises in metrics.\n const resolvedStages: StageHealth[] = await Promise.all(\n stages.map(async (s: any) => {\n const metrics = await Promise.all(s.metrics);\n if (metrics.length === 0) {\n return {\n stage: s.stage,\n label: STAGE_META[s.stage as Stage].label,\n description: STAGE_META[s.stage as Stage].description,\n metrics,\n health_score: 0,\n diagnosis: \"No data — connect integrations for this stage.\",\n };\n }\n\n // ── ABSOLUTE benchmark-based health (replaces broken warm-start baseline scoring) ──\n // Each metric gets a 0-100 score against industry benchmarks. Stage health = mean.\n const absoluteScores: number[] = [];\n const deltaContributions: number[] = [];\n for (const m of metrics as any[]) {\n const abs = normalizeAbsolute(m.integration, m.metric, m.value);\n if (abs !== null) absoluteScores.push(abs);\n // Chronological delta_7d bonus: ±10 added to absolute score if trending.\n if (m.delta_7d != null && abs !== null) {\n const BAD_WHEN_UP = new Set([\"bounce_rate\"]);\n const direction = BAD_WHEN_UP.has(m.metric) ? -1 : 1;\n deltaContributions.push(Math.max(-15, Math.min(15, m.delta_7d * direction * 0.5)));\n }\n }\n const baseHealth = absoluteScores.length\n ? absoluteScores.reduce((a, b) => a + b, 0) / absoluteScores.length\n : 50;\n const deltaBonus = deltaContributions.length\n ? deltaContributions.reduce((a, b) => a + b, 0) / deltaContributions.length\n : 0;\n const health_score = Math.max(0, Math.min(100, Math.round(baseHealth + deltaBonus)));\n\n // Chronological-aware diagnosis: state of absolute health + 7d trend direction.\n const avgTrend = deltaContributions.length\n ? deltaContributions.reduce((a, b) => a + b, 0) / deltaContributions.length\n : 0;\n let diagnosis: string;\n if (health_score >= 75) diagnosis = \"Healthy — metrics meet or exceed SaaS benchmarks.\" + (avgTrend > 2 ? \" Trending up week-over-week.\" : avgTrend < -2 ? \" Mild 7d softening — watch closely.\" : \"\");\n else if (health_score >= 50) diagnosis = \"Moderate — room to improve against benchmarks.\" + (avgTrend > 2 ? \" 7d momentum is positive.\" : avgTrend < -2 ? \" 7d trend is negative.\" : \"\");\n else if (health_score >= 25) diagnosis = \"Underperforming benchmarks — focus optimization here.\" + (avgTrend < -2 ? \" Compounding 7d decline.\" : \"\");\n else diagnosis = \"Critical — metrics far below industry baselines. Prioritize this stage.\";\n\n return {\n stage: s.stage,\n label: STAGE_META[s.stage as Stage].label,\n description: STAGE_META[s.stage as Stage].description,\n metrics,\n health_score,\n diagnosis,\n };\n })\n );\n\n // Bottleneck = lowest-health stage with data.\n const candidates = resolvedStages.filter((s) => s.metrics.length > 0);\n const bottleneckStage = candidates.length\n ? candidates.reduce((lo, s) => (s.health_score < lo.health_score ? s : lo))\n : null;\n\n // Weighted overall_health — only counts stages with data so a missing integration\n // doesn't crater the composite. Weights renormalize when a stage is missing.\n const stageWeights: Record<Stage, number> = { TOFU: 0.25, MOFU: 0.40, BOFU: 0.35 };\n const dataStages = resolvedStages.filter((s) => s.metrics.length > 0);\n const overallHealth = dataStages.length\n ? (() => {\n const weightSum = dataStages.reduce((acc, s) => acc + stageWeights[s.stage], 0);\n const weighted = dataStages.reduce((acc, s) => acc + s.health_score * stageWeights[s.stage], 0);\n return Math.max(0, Math.min(100, weighted / weightSum));\n })()\n : 0;\n\n const summary: FunnelSummary = {\n stages: resolvedStages,\n bottleneck: bottleneckStage\n ? { stage: bottleneckStage.stage, reason: bottleneckStage.diagnosis }\n : null,\n overall_health: Math.round(overallHealth),\n total_metrics_observed: allMetrics.length,\n total_integrations_connected: new Set(allMetrics.map((m) => m.integration)).size,\n generated_at: new Date().toISOString(),\n };\n\n await setCached(cacheKey, summary, 300);\n // Also cache under the legacy key so getCorrelations() (which calls getFunnelSummary() without\n // days) can still find a warm entry when the default 30d summary was just computed.\n if (days === 30) await setCached(\"funnel_summary\", summary, 300);\n return summary;\n}\n\n// ─── CORRELATIONS ────────────────────────────────────────────────────────────\nfunction getMetric(all: MetricPoint[], integration: string, metric: string): MetricPoint | undefined {\n return all.find((m) => m.integration === integration && m.metric === metric);\n}\n\nexport async function getCorrelations(force = false): Promise<CorrelationSignal[]> {\n if (!force) {\n const cached = await getCached(\"correlations\");\n if (cached) return cached;\n }\n const summary = await getFunnelSummary();\n const all = summary.stages.flatMap((s) => s.metrics);\n const signals: CorrelationSignal[] = [];\n\n const dr = getMetric(all, \"ahrefs\", \"domain_rating\");\n const gscClicks = getMetric(all, \"gsc\", \"clicks\");\n const gscCtr = getMetric(all, \"gsc\", \"ctr\");\n const gscImp = getMetric(all, \"gsc\", \"impressions\");\n const sessions = getMetric(all, \"ga4\", \"sessions\");\n const engTime = getMetric(all, \"ga4\", \"engagement_time\");\n const conv = getMetric(all, \"ga4\", \"conversions\");\n const bounce = getMetric(all, \"ga4\", \"bounce_rate\");\n const liImp = getMetric(all, \"linkedin\", \"post_impressions\");\n\n const deltaOf = (m: any) => (m?.delta_pct ?? 0) as number;\n\n // Rule 1: Ahrefs DR up + GSC clicks up\n if (dr && gscClicks && deltaOf(dr) > 0 && deltaOf(gscClicks) > 0) {\n signals.push({\n id: \"dr-clicks\", source_metric: { integration: dr.integration, metric: dr.metric, value: dr.value, unit: dr.unit },\n target_metric: { integration: gscClicks.integration, metric: gscClicks.metric, value: gscClicks.value, unit: gscClicks.unit },\n direction: \"positive\", strength: \"strong\", lag_days: 30,\n narrative: \"Domain authority growth is converting to organic search clicks — SEO investment is compounding.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.78,\n action_prompt: \"Double down on the keyword clusters where rankings improved this month — let the AI prioritize.\",\n });\n }\n // Rule 2: GSC CTR up + sessions up\n if (gscCtr && sessions && deltaOf(gscCtr) > 0 && deltaOf(sessions) > 0) {\n signals.push({\n id: \"ctr-sessions\", source_metric: { integration: gscCtr.integration, metric: gscCtr.metric, value: gscCtr.value, unit: gscCtr.unit },\n target_metric: { integration: sessions.integration, metric: sessions.metric, value: sessions.value, unit: sessions.unit },\n direction: \"positive\", strength: \"strong\", lag_days: 7,\n narrative: \"Search result click-through improvements are driving qualified session growth.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.82,\n action_prompt: \"Identify the top 5 pages whose CTR improved and roll their title/meta patterns to similar pages.\",\n });\n }\n // Rule 3: engagement time up + conversions up\n if (engTime && conv && deltaOf(engTime) > 0 && deltaOf(conv) > 0) {\n signals.push({\n id: \"eng-conv\", source_metric: { integration: engTime.integration, metric: engTime.metric, value: engTime.value, unit: engTime.unit },\n target_metric: { integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit },\n direction: \"positive\", strength: \"moderate\", lag_days: 14,\n narrative: \"Content quality improvements correlate with downstream conversion lift.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.68,\n action_prompt: \"Pin the highest-engagement pages above the fold in nav — they're proven to convert.\",\n });\n }\n // Rule 4: high impressions + low CTR\n if (gscImp && gscCtr && gscImp.value > 1000 && gscCtr.value < 3) {\n signals.push({\n id: \"imp-low-ctr\", source_metric: { integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit },\n target_metric: { integration: gscCtr.integration, metric: gscCtr.metric, value: gscCtr.value, unit: gscCtr.unit },\n direction: \"negative\", strength: \"moderate\", lag_days: 0,\n narrative: \"High impressions with poor CTR indicates title/meta description optimization opportunity.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.85,\n action_prompt: \"Run a Content Weaver agent to rewrite titles + metas for the top 20 high-impression / low-CTR queries.\",\n });\n }\n // Rule 5: bounce up + conversions down (funnel leak)\n if (bounce && conv && deltaOf(bounce) > 5 && deltaOf(conv) < -5) {\n signals.push({\n id: \"bounce-conv-leak\", source_metric: { integration: bounce.integration, metric: bounce.metric, value: bounce.value, unit: bounce.unit },\n target_metric: { integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit },\n direction: \"negative\", strength: \"strong\", lag_days: 0,\n narrative: \"Traffic quality or landing page relevance degrading — funnel leakage at MOFU.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.74,\n action_prompt: \"Pull the top entry pages with rising bounce, audit message-market fit + page speed.\",\n });\n }\n // Rule 6: LinkedIn impressions up — social priming\n if (liImp && deltaOf(liImp) > 10) {\n signals.push({\n id: \"li-warmup\", source_metric: { integration: liImp.integration, metric: liImp.metric, value: liImp.value, unit: liImp.unit },\n target_metric: { integration: \"gmail\", metric: \"open_rate\", value: 0, unit: \"%\" },\n direction: \"positive\", strength: \"moderate\", lag_days: 21,\n narrative: \"Brand visibility on LinkedIn is warming cold email outreach — social priming effect.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.55,\n action_prompt: \"Schedule outreach sequences to land 3-5 days after high-engagement LinkedIn posts.\",\n });\n }\n\n // Additional correlation rules (extends set from 6 → 12).\n const refDom = getMetric(all, \"ahrefs\", \"referring_domains\");\n const posts = getMetric(all, \"wordpress\", \"published_posts\");\n const pps = getMetric(all, \"ga4\", \"pages_per_session\");\n\n // Rule 7: DR ↑ → referring_domains ↑ (lag 14-30d, moderate)\n if (dr && refDom && deltaOf(dr) > 0 && deltaOf(refDom) > 0) {\n signals.push({\n id: \"dr-refdomains\", source_metric: {integration: dr.integration, metric: dr.metric, value: dr.value, unit: dr.unit},\n target_metric: {integration: refDom.integration, metric: refDom.metric, value: refDom.value, unit: refDom.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 21,\n narrative: \"Higher domain authority is attracting new backlinks as more sites notice your ranked content.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.65,\n action_prompt: \"Identify the top 3 pages gaining new backlinks and double down on their content angle.\",\n });\n }\n // Rule 8: referring_domains ↑ → GSC impressions ↑ (lag 14-21d, strong)\n if (refDom && gscImp && deltaOf(refDom) > 0 && deltaOf(gscImp) > 0) {\n signals.push({\n id: \"refdomains-impressions\", source_metric: {integration: refDom.integration, metric: refDom.metric, value: refDom.value, unit: refDom.unit},\n target_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n direction: \"positive\", strength: \"strong\", lag_days: 18,\n narrative: \"New backlinks are pushing more keywords into ranking range, expanding search visibility.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.72,\n action_prompt: \"Pitch guest posts to high-DR sites in the same vertical to amplify the lift.\",\n });\n }\n // Rule 9: GSC impressions ↑ + CTR stable → GSC clicks ↑ (lag 0-7d, strong, near 1:1)\n if (gscImp && gscClicks && deltaOf(gscImp) > 5 && Math.abs(deltaOf(gscCtr) || 0) < 5 && deltaOf(gscClicks) > 0) {\n signals.push({\n id: \"imp-clicks\", source_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n target_metric: {integration: gscClicks.integration, metric: gscClicks.metric, value: gscClicks.value, unit: gscClicks.unit},\n direction: \"positive\", strength: \"strong\", lag_days: 3,\n narrative: \"Impression growth with stable CTR is converting directly to click volume — pure ranking lift.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.80,\n action_prompt: \"Track which queries gained impressions this period and ensure landing pages can absorb the traffic.\",\n });\n }\n // Rule 10: WordPress posts ↑ → GSC impressions ↑ (lag 30-90d, moderate)\n if (posts && gscImp && deltaOf(posts) > 0 && deltaOf(gscImp) > 0) {\n signals.push({\n id: \"posts-impressions\", source_metric: {integration: posts.integration, metric: posts.metric, value: posts.value, unit: posts.unit},\n target_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 60,\n narrative: \"Recent content publishing cadence is now accumulating ranking signals (4-12 week lag).\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.58,\n action_prompt: \"Maintain publishing cadence — backlog of posts will continue to compound impressions over 90 days.\",\n });\n }\n // Rule 11: pages_per_session ↑ → conversions ↑ (lag 7d, moderate)\n if (pps && conv && deltaOf(pps) > 0 && deltaOf(conv) > 0) {\n signals.push({\n id: \"pps-cvr\", source_metric: {integration: pps.integration, metric: pps.metric, value: pps.value, unit: pps.unit},\n target_metric: {integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 7,\n narrative: \"Deeper site exploration correlates with higher purchase intent — content map is working.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.62,\n action_prompt: \"Surface high-engagement secondary pages in primary nav to push more visitors into the explore loop.\",\n });\n }\n // Rule 12: Gmail messages_total trend — proxy for outreach velocity (qualified-interest signal)\n const gmailMsgs = getMetric(all, \"gmail\", \"messages_total\");\n if (gmailMsgs && deltaOf(gmailMsgs) > 10) {\n signals.push({\n id: \"gmail-velocity\", source_metric: {integration: gmailMsgs.integration, metric: gmailMsgs.metric, value: gmailMsgs.value, unit: gmailMsgs.unit},\n target_metric: {integration: \"outlook\", metric: \"meetings_booked\", value: 0, unit: \"mtgs\"},\n direction: \"positive\", strength: \"moderate\", lag_days: 3,\n narrative: \"Outbound email volume is up — reply→meeting conversion typically follows within 2-5 days.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.50,\n action_prompt: \"Block calendar windows next week — meeting requests usually cluster 3-5 days after email volume spikes.\",\n });\n }\n\n await setCached(\"correlations\", signals, 300);\n return signals;\n}\n\n// ─── RECOMMENDATIONS ────────────────────────────────────────────────────────\nexport async function getRecommendations(): Promise<Array<{ title: string; action: string; priority: string }>> {\n const cached = await getCached(\"recommendations\");\n if (cached) return cached;\n\n const summary = await getFunnelSummary();\n const correlations = await getCorrelations();\n const recs: Array<{ title: string; action: string; priority: string }> = [];\n\n // Top recommendation: address bottleneck\n if (summary.bottleneck) {\n recs.push({\n title: `Fix the ${summary.bottleneck.stage} bottleneck`,\n action: `Why is ${summary.bottleneck.stage} health low?`,\n priority: \"critical\",\n });\n }\n // Highest-confidence positive signal: amplify it\n const positives = correlations.filter((c) => c.direction === \"positive\").sort((a, b) => b.confidence - a.confidence);\n if (positives[0]) {\n recs.push({\n title: positives[0].narrative.slice(0, 60) + \"…\",\n action: `How do I amplify the ${positives[0].source_metric.metric} → ${positives[0].target_metric.metric} chain?`,\n priority: \"high\",\n });\n }\n // Highest-confidence negative signal: investigate\n const negatives = correlations.filter((c) => c.direction === \"negative\").sort((a, b) => b.confidence - a.confidence);\n if (negatives[0]) {\n recs.push({\n title: `Investigate: ${negatives[0].narrative.slice(0, 50)}…`,\n action: negatives[0].action_prompt,\n priority: \"high\",\n });\n }\n\n // Fallback if no real data\n if (recs.length === 0) {\n recs.push(\n { title: \"Connect Google Workspace to unlock funnel intelligence\", action: \"How do I connect Google?\", priority: \"high\" },\n { title: \"Connect Ahrefs to track TOFU authority signals\", action: \"What does Ahrefs add to my funnel?\", priority: \"medium\" },\n { title: \"Configure GA4 property ID for MOFU/BOFU visibility\", action: \"How do I set up GA4?\", priority: \"medium\" },\n );\n }\n\n await setCached(\"recommendations\", recs.slice(0, 5), 300);\n return recs.slice(0, 5);\n}\n\n// ─── MQL / SQL LEAD SCORING ──────────────────────────────────────────────────\n// B2B SaaS qualification math:\n// MQL score = sum of behavioral signals (opens, clicks, pricing views, downloads)\n// MQL threshold: ≥ 40\n// SQL = 2+ qualified signals (reply, accepted connection, booked meeting, requested demo)\n// Since Nexus has no CRM, we derive from Gmail inbox threads + Outlook calendar overlap.\nexport interface LeadScore {\n email: string;\n name: string;\n company: string;\n mql_score: number;\n mql_qualified: boolean; // score >= 40\n sql_signals: string[]; // list of qualified signals fired\n sql_qualified: boolean; // signals.length >= 2\n stage: \"cold\" | \"MQL\" | \"SQL\";\n signal_breakdown: Record<string, number>;\n source: \"gmail\" | \"outlook\" | \"linkedin\";\n last_touch_at?: string;\n}\n\nexport async function getLeadScores(): Promise<LeadScore[]> {\n const cached = await getCached(\"lead_scores\");\n if (cached) return cached;\n\n const leads = new Map<string, LeadScore>();\n const getLead = (email: string, name: string, company: string, source: any): LeadScore => {\n if (!leads.has(email)) {\n leads.set(email, {\n email, name, company, source,\n mql_score: 0, mql_qualified: false,\n sql_signals: [], sql_qualified: false,\n stage: \"cold\", signal_breakdown: {},\n });\n }\n return leads.get(email)!;\n };\n\n // ── Pull Outlook inbox: each sender gets points for engagement signals ──\n let outlookCreds = await getCredentials(\"outlook\");\n if (outlookCreds?.access_token) {\n try {\n const r = await fetch(\"https://graph.microsoft.com/v1.0/me/messages?$top=50&$orderby=receivedDateTime desc&$select=from,subject,isRead,body,receivedDateTime\",\n { headers: { Authorization: `Bearer ${outlookCreds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n for (const m of (d.value || [])) {\n const email = m.from?.emailAddress?.address;\n if (!email) continue;\n const name = m.from?.emailAddress?.name || email.split(\"@\")[0];\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = getLead(email, name, company, \"outlook\");\n const subj = (m.subject || \"\").toLowerCase();\n const body = (m.body?.content || \"\").toLowerCase();\n // MQL signals\n if (m.isRead === false) { lead.mql_score += 5; lead.signal_breakdown.unread = (lead.signal_breakdown.unread||0)+5; }\n if (/click|visited|opened/.test(body)) { lead.mql_score += 15; lead.signal_breakdown.clicks_link = 15; }\n if (/pricing|plans|cost/.test(subj+body)) { lead.mql_score += 25; lead.signal_breakdown.visits_pricing = 25; }\n if (/download|whitepaper|ebook|guide/.test(subj+body)) { lead.mql_score += 20; lead.signal_breakdown.downloads_content = 20; }\n if (/webinar|registered/.test(subj+body)) { lead.mql_score += 20; lead.signal_breakdown.attends_webinar = 20; }\n lead.mql_score += 10; lead.signal_breakdown.opens_email_sequence = 10;\n // SQL signals\n if (/re:|reply|response/i.test(subj) && !subj.startsWith(\"re:\")) {\n if (!lead.sql_signals.includes(\"reply_to_outreach_email\")) lead.sql_signals.push(\"reply_to_outreach_email\");\n }\n if (/demo|trial|proposal|quote/.test(subj+body)) {\n if (!lead.sql_signals.includes(\"requested_demo_or_pricing\")) lead.sql_signals.push(\"requested_demo_or_pricing\");\n }\n lead.last_touch_at = m.receivedDateTime || lead.last_touch_at;\n }\n }\n } catch {}\n\n // ── Pull Outlook calendar to detect booked discovery calls ──\n try {\n const start = new Date().toISOString();\n const end = new Date(Date.now() + 30 * 86400000).toISOString();\n const r = await fetch(`https://graph.microsoft.com/v1.0/me/calendar/events?$top=50&$filter=start/dateTime ge '${start}' and start/dateTime le '${end}'&$select=attendees,subject`,\n { headers: { Authorization: `Bearer ${outlookCreds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n for (const ev of (d.value || [])) {\n for (const a of (ev.attendees || [])) {\n const email = a.emailAddress?.address;\n if (!email) continue;\n const name = a.emailAddress?.name || email.split(\"@\")[0];\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = leads.get(email) || getLead(email, name, company, \"outlook\");\n if (!lead.sql_signals.includes(\"booked_discovery_call\")) lead.sql_signals.push(\"booked_discovery_call\");\n }\n }\n }\n } catch {}\n }\n\n // ── (A6) Gmail thread analysis — runs independent of Outlook so Google-only users still get leads.\n // Mirrors Outlook deal-intent scoring against the recent inbox.\n try {\n const gmailCreds = await getCredentials(\"gmail\");\n const token = gmailCreds?.access_token;\n if (token) {\n const listR = await fetch(\"https://gmail.googleapis.com/gmail/v1/users/me/messages?maxResults=30&q=in:inbox+newer_than:30d\",\n { headers: { Authorization: `Bearer ${token}` } });\n if (listR.ok) {\n const list = await safeJson(listR);\n const ids: string[] = (list.messages || []).map((m: any) => m.id).slice(0, 30);\n const fetched = await Promise.allSettled(ids.map((id) =>\n fetch(`https://gmail.googleapis.com/gmail/v1/users/me/messages/${id}?format=metadata&metadataHeaders=From&metadataHeaders=Subject&metadataHeaders=Date`,\n { headers: { Authorization: `Bearer ${token}` } }).then((r) => r.ok ? r.json() : null)\n ));\n for (const f of fetched) {\n if (f.status !== \"fulfilled\" || !f.value) continue;\n const msg: any = f.value;\n const headers: any[] = msg.payload?.headers || [];\n const findH = (n: string) => headers.find((h) => h.name?.toLowerCase() === n.toLowerCase())?.value || \"\";\n const from = findH(\"From\");\n const subject = findH(\"Subject\");\n const date = findH(\"Date\");\n const snippet = (msg.snippet || \"\").toLowerCase();\n const emailMatch = from.match(/<([^>]+)>|([\\w.+-]+@[\\w.-]+)/);\n const email = (emailMatch?.[1] || emailMatch?.[2] || \"\").toLowerCase();\n if (!email) continue;\n const nameMatch = from.match(/^([^<]+)</);\n const name = (nameMatch?.[1] || email.split(\"@\")[0]).trim().replace(/\"/g, \"\");\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = leads.get(email) || getLead(email, name, company, \"gmail\");\n const subj = (subject || \"\").toLowerCase();\n const text = subj + \" \" + snippet;\n // Deal-intent MQL scoring weights\n if (/pricing|plans|cost|how much/.test(text)) {\n lead.mql_score += 25; lead.signal_breakdown.gmail_pricing = (lead.signal_breakdown.gmail_pricing || 0) + 25;\n }\n if (/proposal|quote|contract|sow|statement of work/.test(text)) {\n lead.mql_score += 20; lead.signal_breakdown.gmail_proposal = (lead.signal_breakdown.gmail_proposal || 0) + 20;\n if (!lead.sql_signals.includes(\"requested_proposal\")) lead.sql_signals.push(\"requested_proposal\");\n }\n if (/demo|trial|walkthrough|kickoff/.test(text)) {\n lead.mql_score += 20; lead.signal_breakdown.gmail_demo = (lead.signal_breakdown.gmail_demo || 0) + 20;\n if (!lead.sql_signals.includes(\"requested_demo_or_trial\")) lead.sql_signals.push(\"requested_demo_or_trial\");\n }\n if (/^re:/i.test(subj.trim())) {\n if (!lead.sql_signals.includes(\"replied_to_thread\")) lead.sql_signals.push(\"replied_to_thread\");\n }\n lead.mql_score += 5; lead.signal_breakdown.gmail_touch = (lead.signal_breakdown.gmail_touch || 0) + 5;\n if (date) {\n const dt = new Date(date).toISOString();\n if (!lead.last_touch_at || dt > lead.last_touch_at) lead.last_touch_at = dt;\n }\n }\n }\n }\n } catch { /* gmail optional */ }\n\n // Finalize: classify each lead\n const result: LeadScore[] = [];\n for (const lead of leads.values()) {\n lead.mql_qualified = lead.mql_score >= 40;\n lead.sql_qualified = lead.sql_signals.length >= 2;\n lead.stage = lead.sql_qualified ? \"SQL\" : lead.mql_qualified ? \"MQL\" : \"cold\";\n result.push(lead);\n }\n // Record state transitions so engagement velocity has data to compute from.\n // recordLeadStateChange skips inserts when the state hasn't changed since last write.\n try {\n const { recordLeadStateChange } = await import(\"./aio.ts\");\n await Promise.allSettled(result.map((l) => recordLeadStateChange(l.email, l.stage)));\n } catch { /* non-critical */ }\n // Sort: SQL first, then MQL, then by score desc\n result.sort((a, b) => {\n const rank = (l: LeadScore) => l.sql_qualified ? 2 : l.mql_qualified ? 1 : 0;\n return (rank(b) - rank(a)) || (b.mql_score - a.mql_score);\n });\n\n const sliced = result.slice(0, 50);\n await setCached(\"lead_scores\", sliced, 300);\n return sliced;\n}\n\n// ─── AIO MARKETING KPIs (the \"math problem\" surface) ────────────────────────\n// CAC = est_marketing_spend / new_customers_acquired\n// LTV = avg_deal_size × retention_months (default retention_months = 12 — configurable)\n// LTV/CAC ratio: healthy SaaS benchmark is ≥ 3:1\n// Funnel conversion rates: TOFU→MOFU (sessions/clicks), MOFU→BOFU (conversions/sessions)\n// mer = Marketing Efficiency Ratio (revenue / spend). Renamed from roi_multiplier. Healthy bench ≥ 3.0.\n// pipeline_coverage = pipeline_value / sales_quota. Healthy ≥ 3.0 (3x quota in pipe).\nexport interface AIOKpis {\n cac: number | null; // Customer Acquisition Cost\n ltv: number | null; // Lifetime Value\n ltv_cac_ratio: number | null; // Higher = better unit economics\n mer: number | null; // Marketing Efficiency Ratio = revenue / marketing spend\n roi_multiplier: number | null; // DEPRECATED alias of mer — kept for backwards compat\n pipeline_coverage: number | null; // pipeline_value / sales_quota\n pipeline_value: number;\n sales_quota: number;\n funnel_conv_tofu_mofu: number | null; // sessions / clicks (%)\n funnel_conv_mofu_bofu: number | null; // conversions / sessions (%)\n conversions: number;\n sessions: number;\n clicks: number;\n marketing_spend: number;\n avg_deal_size: number;\n retention_months: number;\n benchmark_status: {\n ltv_cac: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n mer: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n pipeline_coverage: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n };\n generated_at: string;\n}\n\nasync function getAppConfigNumber(key: string): Promise<number> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [key]);\n if (!r.rows?.length) return 0;\n const v = String((r.rows[0] as any[])[0]).replace(/^\"|\"$/g, \"\");\n return parseFloat(v) || 0;\n } catch { return 0; }\n}\n\nexport async function getAIOKpis(): Promise<AIOKpis> {\n const cached = await getCached(\"aio_kpis\");\n if (cached) return cached;\n\n const summary = await getFunnelSummary();\n const all = summary.stages.flatMap((s) => s.metrics);\n const sessions = getMetric(all, \"ga4\", \"sessions\")?.value || 0;\n const conversions = getMetric(all, \"ga4\", \"conversions\")?.value || 0;\n const clicks = getMetric(all, \"gsc\", \"clicks\")?.value || 0;\n\n const avg_deal_size = await getAppConfigNumber(\"avg_deal_size\");\n const marketing_spend = await getAppConfigNumber(\"est_marketing_spend\");\n const retention_months = (await getAppConfigNumber(\"retention_months\")) || 12;\n const pipeline_value = await getAppConfigNumber(\"pipeline_value\");\n const sales_quota = await getAppConfigNumber(\"sales_quota\");\n\n // Real math, with explicit null when inputs missing.\n const cac = (marketing_spend > 0 && conversions > 0) ? marketing_spend / conversions : null;\n const ltv = avg_deal_size > 0 ? avg_deal_size * retention_months : null;\n const ltv_cac_ratio = (ltv && cac && cac > 0) ? ltv / cac : null;\n // Marketing Efficiency Ratio = revenue / spend. Healthy bench ≥ 3.0 (Klaviyo/Recharge label).\n const mer = (marketing_spend > 0 && conversions > 0 && avg_deal_size > 0)\n ? (conversions * avg_deal_size) / marketing_spend : null;\n const roi_multiplier = mer; // deprecated alias\n // Pipeline Coverage Ratio — sales-team forward-looking signal.\n const pipeline_coverage = (sales_quota > 0 && pipeline_value > 0) ? pipeline_value / sales_quota : null;\n // QA fix: was producing >100% (e.g. 902%) when denominator was very small. Guard with\n // a minimum-sample floor and clamp to 0-100. Anything outside that range means our\n // metric units don't align (clicks vs sessions vs visits), not a real conversion rate.\n const MIN_SAMPLE = 50;\n const funnel_conv_tofu_mofu = (clicks >= MIN_SAMPLE)\n ? Math.max(0, Math.min(100, (sessions / clicks) * 100))\n : null;\n const funnel_conv_mofu_bofu = (sessions >= MIN_SAMPLE)\n ? Math.max(0, Math.min(100, (conversions / sessions) * 100))\n : null;\n\n const benchStatus = (v: number | null, healthy: number, borderline: number) =>\n v == null ? \"unknown\" as const :\n v >= healthy ? \"healthy\" as const :\n v >= borderline ? \"borderline\" as const :\n \"unhealthy\" as const;\n const ltv_cac_status = benchStatus(ltv_cac_ratio, 3, 1);\n const mer_status = benchStatus(mer, 3, 1);\n const pipeline_coverage_status = benchStatus(pipeline_coverage, 3, 1.5);\n\n const out: AIOKpis = {\n cac, ltv, ltv_cac_ratio, mer, roi_multiplier,\n pipeline_coverage, pipeline_value, sales_quota,\n funnel_conv_tofu_mofu, funnel_conv_mofu_bofu,\n conversions, sessions, clicks, marketing_spend, avg_deal_size, retention_months,\n benchmark_status: { ltv_cac: ltv_cac_status, mer: mer_status, pipeline_coverage: pipeline_coverage_status },\n generated_at: new Date().toISOString(),\n };\n await setCached(\"aio_kpis\", out, 300);\n return out;\n}\n\n// ─── INTELLIGENCE FEED (LLM-generated narrative cards) ──────────────────────\nexport async function getIntelligenceFeed(): Promise<IntelligenceCard[]> {\n // Cache invalidation — only serve cached feed if the funnel summary it was built\n // from is not newer than the cache. When the user hits ?force=1 on funnel-summary the\n // upstream cache regenerates, and the feed should follow on its next read instead of\n // serving up-to-5-minute-stale narrative cards.\n const cached = await getCachedTimestamp(\"feed\");\n const summaryMeta = await getCachedTimestamp(\"funnel_summary\");\n if (cached && (!summaryMeta || summaryMeta.generated_at <= cached.generated_at)) {\n return cached.payload;\n }\n\n const summary = await getFunnelSummary();\n const correlations = await getCorrelations();\n\n // Deterministic baseline cards from correlations (no LLM call required)\n const baseCards: IntelligenceCard[] = correlations.slice(0, 4).map((c) => ({\n type: c.direction === \"positive\" ? \"win\" : c.direction === \"negative\" ? \"risk\" : \"correlation\",\n priority: c.strength === \"strong\" ? \"critical\" : c.strength === \"moderate\" ? \"high\" : \"medium\",\n title: c.narrative.split(\".\")[0],\n body: `${c.narrative} Lag: ${c.lag_days}d · confidence: ${Math.round(c.confidence * 100)}%`,\n metrics: [\n `${c.source_metric.integration} ${c.source_metric.metric}: ${c.source_metric.value.toFixed(1)}${c.source_metric.unit}`,\n `${c.target_metric.integration} ${c.target_metric.metric}: ${c.target_metric.value.toFixed(1)}${c.target_metric.unit}`,\n ],\n funnel_stages: c.funnel_stage_from === c.funnel_stage_to ? [c.funnel_stage_from] : [c.funnel_stage_from, c.funnel_stage_to],\n agent_name: \"Signal Monitor\",\n timestamp: new Date().toISOString(),\n action_label: \"Investigate →\",\n action_payload: { question: c.action_prompt, context: { signal_id: c.id, narrative: c.narrative } },\n }));\n\n // Add a bottleneck card\n if (summary.bottleneck) {\n baseCards.unshift({\n type: \"anomaly\", priority: \"critical\",\n title: `${summary.bottleneck.stage} stage is the current bottleneck`,\n body: summary.bottleneck.reason,\n metrics: summary.stages\n .find((s) => s.stage === summary.bottleneck!.stage)!\n .metrics.slice(0, 3)\n .map((m) => `${m.integration} ${m.metric}: ${m.value.toFixed(1)}${m.unit}`),\n funnel_stages: [summary.bottleneck.stage],\n agent_name: \"Funnel Sentinel\",\n timestamp: new Date().toISOString(),\n action_label: \"Diagnose →\",\n action_payload: { question: `What's causing the drop in ${summary.bottleneck.stage} health?` },\n });\n }\n\n // Optionally enrich with LLM narrative when an LLM is configured.\n try {\n const cfg = await getLLMConfig();\n if (cfg.api_key && baseCards.length > 0) {\n const ctx = JSON.stringify({\n stages: summary.stages.map((s) => ({ stage: s.stage, health: s.health_score, metrics: s.metrics.length })),\n correlations: correlations.map((c) => ({ from: c.source_metric.metric, to: c.target_metric.metric, dir: c.direction, narrative: c.narrative })),\n });\n const resp = await callLLM(cfg, [{ role: \"user\", content: `Given this funnel state, write ONE concise 2-sentence \"opportunity\" card body (no preamble, no markdown):\\n${ctx}` }], { maxTokens: 200 });\n if (resp.content) {\n baseCards.splice(1, 0, {\n type: \"opportunity\", priority: \"high\",\n title: \"AI Insight: Cross-stage opportunity detected\",\n body: resp.content.trim(),\n metrics: [\"LLM-derived from funnel state\"],\n funnel_stages: [\"TOFU\", \"MOFU\", \"BOFU\"],\n agent_name: \"Nexus Analyst\",\n timestamp: new Date().toISOString(),\n action_label: \"Explore →\",\n action_payload: { question: \"Tell me more about this opportunity and how to act on it.\" },\n });\n }\n }\n } catch {}\n\n // Empty-state placeholder when nothing connected\n if (baseCards.length === 0) {\n baseCards.push({\n type: \"opportunity\", priority: \"medium\",\n title: \"Connect your first integration to unlock intelligence\",\n body: \"Once Ahrefs, GA4, or Search Console is connected, Nexus will start computing causal signals across your funnel.\",\n metrics: [], funnel_stages: [\"TOFU\"], agent_name: \"Onboarding\",\n timestamp: new Date().toISOString(),\n action_label: \"Open Settings →\",\n action_payload: { question: \"Walk me through connecting my first integration.\" },\n });\n }\n\n await setCached(\"feed\", baseCards, 300);\n return baseCards;\n}\n",[],[365,811],"6bb20ca8-5b48-11f1-9649-ee650bb23af1","integrations.ts","317116ec-56ca-11f1-b821-ee650bb23af1",497,["D",1780050202769],27,"// Integration dispatcher. Every per-integration handler lives in its own file under api/integrations/.\n// This file:\n// 1. exposes `getCredentials` to the router for the OAuth callback\n// 2. dispatches /api/integrations/:name/data to the appropriate per-integration handler\n// 3. handles cross-cutting concerns: Google OAuth fallback, auto-refresh on 401, credential guards\nimport { json, safeJson, getCredentials, withGoogleFallback, withMicrosoftFallback, refreshGoogleToken, refreshMicrosoftToken, GOOGLE_SERVICES, MICROSOFT_SERVICES, MICROSOFT_SCOPES, googleClientId, googleClientSecret, googleRedirectUri, IntegrationCredentials } from \"./integrations/_shared.ts\";\n\nimport * as ahrefs from \"./integrations/ahrefs.ts\";\nimport * as ga4 from \"./integrations/ga4.ts\";\nimport * as gsc from \"./integrations/gsc.ts\";\nimport * as youtube from \"./integrations/youtube.ts\";\nimport * as drive from \"./integrations/drive.ts\";\nimport * as sheets from \"./integrations/sheets.ts\";\nimport * as docs from \"./integrations/docs.ts\";\nimport * as calendar from \"./integrations/calendar.ts\";\nimport * as gmail from \"./integrations/gmail.ts\";\nimport * as linkedin from \"./integrations/linkedin.ts\";\nimport * as outlook from \"./integrations/outlook.ts\";\nimport * as wordpress from \"./integrations/wordpress.ts\";\nimport * as producthunt from \"./integrations/producthunt.ts\";\nimport * as onedrive from \"./integrations/onedrive.ts\";\nimport * as teams from \"./integrations/teams.ts\";\nimport * as excel from \"./integrations/excel.ts\";\n\n// Re-export getCredentials so router.ts (OAuth callback) keeps working without changes.\n// GOOGLE_SERVICES is re-exported so the router can resolve shared Google creds before dispatch.\n// google* helpers are re-exported so the router's OAuth handlers read env-first app credentials.\nexport { getCredentials, GOOGLE_SERVICES, MICROSOFT_SCOPES, googleClientId, googleClientSecret, googleRedirectUri };\nexport type { IntegrationCredentials };\n\n// Map of integration name → handler module. Adding a new integration = adding one file + one entry here.\nconst HANDLERS = {\n ahrefs, ga4, gsc, youtube, drive, sheets, docs, calendar, gmail, linkedin, outlook, wordpress, producthunt,\n onedrive, teams, excel,\n} as const;\n\ntype IntegrationName = keyof typeof HANDLERS;\n\n// Built-in test-connection for integrations whose module didn't export their own `test`.\nasync function defaultTestConnection(name: string, creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n // Google OAuth tokeninfo check + per-service scope coverage report.\n if (name === \"google\" || GOOGLE_SERVICES.includes(name)) {\n const res = await fetch(`https://www.googleapis.com/oauth2/v1/tokeninfo?access_token=${encodeURIComponent(creds.access_token ?? \"\")}`);\n if (!res.ok) return { ok: false, message: \"Invalid or expired access token — click Re-authorize Google\" };\n const info = await safeJson(res);\n const scope = (info.scope ?? \"\") as string;\n // Per-service required scopes (any match = granted). Write-capable scopes checked specifically.\n const SERVICES: Array<[string, string[]]> = [\n [\"GA4\", [\"analytics\"]],\n [\"GSC\", [\"webmasters\", \"search-console\"]],\n [\"YouTube\", [\"youtube\"]],\n [\"Drive RW\", [\"auth/drive \"]],\n [\"Sheets RW\", [\"auth/spreadsheets \"]],\n [\"Docs RW\", [\"auth/documents \"]],\n [\"Calendar\", [\"auth/calendar \"]],\n [\"Gmail RD\", [\"gmail.readonly\", \"gmail.modify\"]],\n [\"Gmail SEND\", [\"gmail.send\"]],\n ];\n const scopeWithTrailingSpace = scope + \" \";\n const granted = SERVICES.filter(([, ks]) => ks.some(k => scopeWithTrailingSpace.includes(k))).map(([n]) => n);\n const missing = SERVICES.filter(([, ks]) => !ks.some(k => scopeWithTrailingSpace.includes(k))).map(([n]) => n);\n const msg = missing.length === 0\n ? `Google verified · all ${SERVICES.length} services + write authorized`\n : `Google verified · ${granted.length}/${SERVICES.length}: ${granted.join(\", \")} · MISSING: ${missing.join(\", \")}`;\n return { ok: missing.length === 0, message: msg };\n }\n return { ok: false, message: \"No test implemented for this integration\" };\n}\n\nexport async function handleIntegrationProxy(\n req: Request,\n name: string,\n searchParams: URLSearchParams,\n creds: IntegrationCredentials\n): Promise<Response> {\n const endpoint = searchParams.get(\"endpoint\") ?? \"\";\n\n // Google services share central OAuth — merge in shared token before dispatching.\n if (GOOGLE_SERVICES.includes(name)) creds = await withGoogleFallback(creds);\n // Microsoft Graph services (onedrive/teams/excel) share the central Outlook token.\n if (MICROSOFT_SERVICES.includes(name)) creds = await withMicrosoftFallback(creds);\n\n // Credential guards\n if ([\"ahrefs\", \"producthunt\"].includes(name) && !creds.api_key && !creds.access_token) {\n return json({ error: `${name} credentials not configured. Add them in Settings.` }, 401);\n }\n if (name === \"youtube\" && !creds.api_key && !creds.access_token) {\n return json({ error: \"YouTube requires an API Key or Google OAuth access token. Configure Google in Settings.\" }, 401);\n }\n if ([\"ga4\", \"gsc\", \"drive\", \"sheets\", \"calendar\", \"gmail\", \"docs\"].includes(name) && !creds.access_token) {\n return json({ error: `${name} requires a Google OAuth access token. Connect with Google in Settings.` }, 401);\n }\n if ([\"linkedin\", \"outlook\", \"onedrive\", \"teams\", \"excel\"].includes(name) && !creds.access_token) {\n const hint = MICROSOFT_SERVICES.includes(name) ? \"Connect Outlook (Microsoft) in Settings.\" : \"Add it in Settings.\";\n return json({ error: `${name} access token not configured. ${hint}` }, 401);\n }\n if (name === \"wordpress\" && !creds.api_key) {\n return json({ error: \"WordPress username not configured. Add credentials in Settings.\" }, 401);\n }\n\n // Test endpoint: prefer integration's own `test` export, fall back to default.\n if (endpoint === \"test\") {\n const handler = HANDLERS[name as IntegrationName];\n const tester = (handler as any)?.test;\n const result = tester ? await tester(creds) : await defaultTestConnection(name, creds);\n return json(result, result.ok ? 200 : 401);\n }\n\n // Google = the credential record itself, no API surface of its own beyond the test endpoint.\n if (name === \"google\") return json({ ok: true, message: \"Google OAuth configured\" });\n\n // Build forwarded query string — exclude our own meta params.\n const excluded = new Set([\"endpoint\", \"integration\"]);\n const forwardedParts: string[] = [];\n searchParams.forEach((value, key) => {\n if (!excluded.has(key)) forwardedParts.push(`${encodeURIComponent(key)}=${encodeURIComponent(value)}`);\n });\n const params = forwardedParts.join(\"&\");\n\n const handler = HANDLERS[name as IntegrationName];\n if (!handler) return json({ error: `Unknown integration: ${name}` }, 400);\n\n // Product Hunt's handler signature takes the URLSearchParams directly (uses GraphQL query param).\n const dispatch = async (c: IntegrationCredentials): Promise<Response> => {\n if (name === \"producthunt\") return (handler as any).handle(req, endpoint, searchParams, c);\n return (handler as any).handle(req, endpoint, params, c);\n };\n\n let resp = await dispatch(creds);\n\n // Auto-refresh Google OAuth on 401, retry once.\n if (resp.status === 401 && GOOGLE_SERVICES.includes(name) && creds.refresh_token) {\n const newToken = await refreshGoogleToken();\n if (newToken) {\n creds = { ...creds, access_token: newToken };\n resp = await dispatch(creds);\n }\n }\n\n // Auto-refresh Microsoft Graph OAuth (outlook + onedrive/teams/excel) on 401, retry once.\n if (resp.status === 401 && (name === \"outlook\" || MICROSOFT_SERVICES.includes(name)) && creds.refresh_token) {\n const newToken = await refreshMicrosoftToken();\n if (newToken) {\n creds = { ...creds, access_token: newToken };\n resp = await dispatch(creds);\n }\n }\n\n return resp;\n}\n",[],[365,820],"66e8d938-5916-11f1-aea8-ee650bb23af1","graphify.ts","66e8da1e-5916-11f1-aea8-ee650bb23af1",288,["D",1779808817573],"export async function getGraphify(): Promise<Record<string, unknown>> {\n return {\n version: \"1.0\",\n app: \"Nexus AI Command Center\",\n graph: {\n pages: [\n { id: \"dashboard\", label: \"Command Center\", component: \"CommandCenter\", routes: [\"/api/intelligence/feed\", \"/api/nsm\", \"/api/aio/cohort-summary\"] },\n { id: \"marketing\", label: \"Strategy\", component: \"MarketingWorkspace\", routes: [\"/api/integrations/ahrefs/data\", \"/api/integrations/ga4/data\", \"/api/integrations/gsc/data\"] },\n { id: \"sales\", label: \"Signals\", component: \"SalesWorkspace\", routes: [\"/api/signals\", \"/api/signals/hot-leads\"] },\n { id: \"workspace\", label: \"Team Hub\", component: \"WorkspacePage\", routes: [] },\n { id: \"objectives\", label: \"OKRs\", component: \"ObjectivesPage\", routes: [\"/api/okrs\", \"/api/okrs/dashboard\"] },\n { id: \"calendar\", label: \"Command Calendar\", component: \"CalendarPage\", routes: [\"/api/calendar/unified\"] },\n { id: \"experiments\", label: \"A/B Tests\", component: \"ExperimentBoard\", routes: [\"/api/experiments\"] },\n { id: \"analytics\", label: \"Analytics\", component: \"AnalyticsPage\", routes: [] },\n { id: \"agents\", label: \"Agent Studio\", component: \"AgentStudio\", routes: [\"/api/agents\"] },\n { id: \"settings\", label: \"Settings\", component: \"SettingsPage\", routes: [\"/api/llm/config\", \"/api/integrations\"] },\n ],\n routes: [\n { method: \"GET\", path: \"/api/health\" },\n { method: \"GET\", path: \"/api/config\" },\n { method: \"GET\", path: \"/api/graphify\" },\n { method: \"GET|POST\", path: \"/api/chat\" },\n { method: \"GET\", path: \"/api/llm/config\" },\n { method: \"POST\", path: \"/api/llm/config\" },\n { method: \"GET\", path: \"/api/agents\" },\n { method: \"POST\", path: \"/api/agents\" },\n { method: \"PUT\", path: \"/api/agents/:id\" },\n { method: \"DELETE\", path: \"/api/agents/:id\" },\n { method: \"GET\", path: \"/api/agents/:id/runs\" },\n { method: \"POST\", path: \"/api/agents/:id/runs\" },\n { method: \"GET\", path: \"/api/integrations\" },\n { method: \"POST\", path: \"/api/integrations/:name\" },\n { method: \"DELETE\", path: \"/api/integrations/:name\" },\n { method: \"GET|POST\", path: \"/api/integrations/:name/data\" },\n { method: \"GET\", path: \"/api/intelligence/funnel-summary\" },\n { method: \"GET\", path: \"/api/intelligence/correlations\" },\n { method: \"GET\", path: \"/api/intelligence/recommendations\" },\n { method: \"GET\", path: \"/api/intelligence/leads\" },\n { method: \"GET\", path: \"/api/intelligence/kpis\" },\n { method: \"GET\", path: \"/api/intelligence/feed\" },\n { method: \"GET|POST\", path: \"/api/signals\" },\n { method: \"PUT\", path: \"/api/signals/:id/action\" },\n { method: \"GET\", path: \"/api/signals/hot-leads\" },\n { method: \"GET\", path: \"/api/signals/hot-leads/count\" },\n { method: \"GET|POST\", path: \"/api/okrs\" },\n { method: \"GET\", path: \"/api/okrs/dashboard\" },\n { method: \"PUT|DELETE\", path: \"/api/okrs/:id\" },\n { method: \"POST\", path: \"/api/okrs/:id/key-results\" },\n { method: \"PUT\", path: \"/api/okrs/key-results/:id\" },\n { method: \"GET|POST\", path: \"/api/nsm\" },\n { method: \"POST\", path: \"/api/nsm/record\" },\n { method: \"GET\", path: \"/api/nsm/progress\" },\n { method: \"GET|POST\", path: \"/api/aio/cohorts\" },\n { method: \"DELETE\", path: \"/api/aio/cohorts/:id\" },\n { method: \"GET\", path: \"/api/aio/cohort-summary\" },\n { method: \"GET\", path: \"/api/aio/engagement-velocity\" },\n { method: \"GET\", path: \"/api/digest\" },\n { method: \"POST\", path: \"/api/digest/generate\" },\n { method: \"GET\", path: \"/api/digest/history\" },\n { method: \"GET|POST\", path: \"/api/experiments\" },\n { method: \"GET\", path: \"/api/calendar/unified\" },\n { method: \"GET\", path: \"/api/tasks\" },\n ],\n integrations: [\n \"google\", \"ahrefs\", \"ga4\", \"gsc\", \"gmail\", \"drive\", \"sheets\",\n \"docs\", \"calendar\", \"linkedin\", \"outlook\", \"wordpress\", \"youtube\", \"producthunt\",\n ],\n agents_schema: {\n fields: [\"id\", \"name\", \"workspace\", \"description\", \"system_prompt\", \"tools\", \"model\", \"enabled\", \"role\"],\n workspaces: [\"marketing\", \"sales\"],\n tool_ids: [\"ahrefs\", \"gsc\", \"ga4\", \"wordpress\", \"youtube\", \"linkedin\", \"outlook\", \"gmail\", \"drive\", \"sheets\", \"docs\", \"producthunt\"],\n },\n features: [\n \"AI chat with tool-calling across integrations\",\n \"Intelligence funnel (TOFU/MOFU/BOFU health scoring)\",\n \"North Star Metric tracking with history\",\n \"OKR management with key results\",\n \"ICE-prioritized A/B experiment board\",\n \"Intent signal detection and hot-lead ranking\",\n \"Cohort LTV analysis\",\n \"Weekly growth digest generation\",\n \"Unified calendar (tasks + experiments + objectives + Google Calendar)\",\n \"Multi-provider LLM config (Anthropic, OpenAI, Gemini, custom)\",\n \"Agent Studio with scheduled runs and budget tracking\",\n \"Command Center terminal with AI-native queries\",\n ],\n },\n };\n}\n",[],[365,829],"201ea062-5d69-11f1-ad33-1607ee4eb77e","experiments.ts","973fc862-589e-11f1-8c3f-ee650bb23af1",611,["D",1780284151688],"// Experiments — Phase 1 Framework Layer.\n// ICE-prioritized experiments mapped to funnel stage + AARRR lever.\n// LLM-assisted ICE scoring uses the funnel summary as live context.\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\nimport { propagateExperimentComplete } from \"./propagate.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type FunnelStage = \"TOFU\" | \"MOFU\" | \"BOFU\";\nexport type AARRRLever =\n | \"acquisition\" | \"activation\" | \"retention\" | \"referral\" | \"revenue\";\n\nexport interface Experiment {\n id: string;\n title: string;\n hypothesis: string | null;\n objective_id: string | null;\n funnel_stage: FunnelStage | null;\n aarrr_lever: AARRRLever | null;\n status: string; // idea | running | completed | killed\n impact: number; // 1..10\n confidence: number; // 1..10\n ease: number; // 1..10\n ice_score: number; // app-computed mean of impact/confidence/ease\n owner: string | null;\n start_date: string | null;\n end_date: string | null;\n channel: string | null;\n success_metric: string | null;\n baseline_value: number | null;\n result_value: number | null;\n outcome: string | null; // win | loss | inconclusive\n learnings: string | null;\n uplift_pct: number | null; // app-computed\n created_at: string;\n updated_at: string;\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp10 = (n: any) => {\n const v = Number(n);\n if (!Number.isFinite(v)) return 5;\n return Math.max(1, Math.min(10, Math.round(v)));\n};\n\nconst computeIce = (impact: number, confidence: number, ease: number) =>\n Math.round(((impact + confidence + ease) / 3) * 100) / 100;\n\nfunction computeUplift(baseline: number | null, result: number | null): number | null {\n if (baseline === null || result === null) return null;\n const b = Number(baseline);\n const r = Number(result);\n if (!Number.isFinite(b) || !Number.isFinite(r) || b === 0) return null;\n return Math.round(((r - b) / Math.abs(b)) * 10000) / 100; // 2 decimal pct\n}\n\nfunction rowToExperiment(columns: string[], row: any[]): Experiment {\n const obj = rowToObj(columns, row);\n const impact = clamp10(obj.impact);\n const confidence = clamp10(obj.confidence);\n const ease = clamp10(obj.ease);\n const baseline = obj.baseline_value != null ? Number(obj.baseline_value) : null;\n const result = obj.result_value != null ? Number(obj.result_value) : null;\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n hypothesis: obj.hypothesis ?? null,\n objective_id: obj.objective_id ?? null,\n funnel_stage: (obj.funnel_stage ?? null) as FunnelStage | null,\n aarrr_lever: (obj.aarrr_lever ?? null) as AARRRLever | null,\n status: String(obj.status ?? \"idea\"),\n impact,\n confidence,\n ease,\n ice_score: computeIce(impact, confidence, ease),\n owner: obj.owner ?? null,\n start_date: obj.start_date ?? null,\n end_date: obj.end_date ?? null,\n channel: obj.channel ?? null,\n success_metric: obj.success_metric ?? null,\n baseline_value: baseline,\n result_value: result,\n outcome: obj.outcome ?? null,\n learnings: obj.learnings ?? null,\n uplift_pct: computeUplift(baseline, result),\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n };\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listExperiments(filters: {\n status?: string;\n funnel_stage?: string;\n aarrr_lever?: string;\n} = {}): Promise<Experiment[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.status) { where.push(\"status = ?\"); args.push(filters.status); }\n if (filters.funnel_stage) { where.push(\"funnel_stage = ?\"); args.push(filters.funnel_stage); }\n if (filters.aarrr_lever) { where.push(\"aarrr_lever = ?\"); args.push(filters.aarrr_lever); }\n const sql = `SELECT id, title, hypothesis, objective_id, funnel_stage, aarrr_lever,\n status, impact, confidence, ease, owner, start_date, end_date,\n channel, success_metric, baseline_value, result_value, outcome,\n learnings, created_at, updated_at\n FROM experiments\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}`;\n const r = await sqlite.execute(sql, args);\n const experiments = (r.rows || []).map(\n (row) => rowToExperiment(r.columns, row as any[]),\n );\n // Sort by ICE DESC in app code.\n experiments.sort((a, b) => b.ice_score - a.ice_score);\n return experiments;\n } catch {\n return [];\n }\n}\n\nasync function getExperimentById(id: string): Promise<Experiment | null> {\n const r = await sqlite.execute(\n `SELECT id, title, hypothesis, objective_id, funnel_stage, aarrr_lever,\n status, impact, confidence, ease, owner, start_date, end_date,\n channel, success_metric, baseline_value, result_value, outcome,\n learnings, created_at, updated_at\n FROM experiments WHERE id = ?`,\n [id],\n );\n if (!r.rows?.length) return null;\n return rowToExperiment(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Write ───────────────────────────────────────────────────────────────────\n\nexport async function createExperiment(\n data: Partial<Experiment>,\n): Promise<Experiment> {\n const now = new Date().toISOString();\n const impact = clamp10(data.impact ?? 5);\n const confidence = clamp10(data.confidence ?? 5);\n const ease = clamp10(data.ease ?? 5);\n\n await sqlite.execute(\n `INSERT INTO experiments\n (title, hypothesis, objective_id, funnel_stage, aarrr_lever, status,\n impact, confidence, ease, owner, start_date, end_date, channel,\n success_metric, baseline_value, result_value, outcome, learnings,\n created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled Experiment\",\n data.hypothesis ?? null,\n data.objective_id ?? null,\n data.funnel_stage ?? null,\n data.aarrr_lever ?? null,\n data.status ?? \"idea\",\n impact, confidence, ease,\n data.owner ?? null,\n data.start_date ?? null,\n data.end_date ?? null,\n data.channel ?? null,\n data.success_metric ?? null,\n data.baseline_value ?? null,\n data.result_value ?? null,\n data.outcome ?? null,\n data.learnings ?? null,\n now, now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return (await getExperimentById(id))!;\n}\n\nexport async function updateExperiment(\n id: string,\n data: Partial<Experiment>,\n): Promise<Experiment | null> {\n try {\n const cur = await getExperimentById(id);\n if (!cur) return null;\n const now = new Date().toISOString();\n const next = {\n title: data.title ?? cur.title,\n hypothesis: \"hypothesis\" in data ? (data.hypothesis ?? null) : cur.hypothesis,\n objective_id: \"objective_id\" in data ? (data.objective_id ?? null) : cur.objective_id,\n funnel_stage: \"funnel_stage\" in data ? (data.funnel_stage ?? null) : cur.funnel_stage,\n aarrr_lever: \"aarrr_lever\" in data ? (data.aarrr_lever ?? null) : cur.aarrr_lever,\n status: data.status ?? cur.status,\n impact: data.impact != null ? clamp10(data.impact) : cur.impact,\n confidence: data.confidence != null ? clamp10(data.confidence) : cur.confidence,\n ease: data.ease != null ? clamp10(data.ease) : cur.ease,\n owner: \"owner\" in data ? (data.owner ?? null) : cur.owner,\n start_date: \"start_date\" in data ? (data.start_date ?? null) : cur.start_date,\n end_date: \"end_date\" in data ? (data.end_date ?? null) : cur.end_date,\n channel: \"channel\" in data ? (data.channel ?? null) : cur.channel,\n success_metric: \"success_metric\" in data ? (data.success_metric ?? null) : cur.success_metric,\n baseline_value: \"baseline_value\" in data ? (data.baseline_value ?? null) : cur.baseline_value,\n result_value: \"result_value\" in data ? (data.result_value ?? null) : cur.result_value,\n outcome: \"outcome\" in data ? (data.outcome ?? null) : cur.outcome,\n learnings: \"learnings\" in data ? (data.learnings ?? null) : cur.learnings,\n };\n await sqlite.execute(\n `UPDATE experiments SET\n title = ?, hypothesis = ?, objective_id = ?, funnel_stage = ?, aarrr_lever = ?,\n status = ?, impact = ?, confidence = ?, ease = ?, owner = ?, start_date = ?,\n end_date = ?, channel = ?, success_metric = ?, baseline_value = ?,\n result_value = ?, outcome = ?, learnings = ?, updated_at = ?\n WHERE id = ?`,\n [\n next.title, next.hypothesis, next.objective_id, next.funnel_stage, next.aarrr_lever,\n next.status, next.impact, next.confidence, next.ease, next.owner, next.start_date,\n next.end_date, next.channel, next.success_metric, next.baseline_value,\n next.result_value, next.outcome, next.learnings, now, id,\n ],\n );\n return await getExperimentById(id);\n } catch {\n return null;\n }\n}\n\nexport async function deleteExperiment(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM experiments WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\nexport async function runExperiment(\n id: string,\n baseline_value: number,\n): Promise<Experiment | null> {\n try {\n const now = new Date().toISOString();\n await sqlite.execute(\n `UPDATE experiments\n SET status = 'running', start_date = ?, baseline_value = ?, updated_at = ?\n WHERE id = ?`,\n [now, Number(baseline_value) || 0, now, id],\n );\n return await getExperimentById(id);\n } catch {\n return null;\n }\n}\n\nexport async function completeExperiment(\n id: string,\n data: { result_value: number; outcome: string; learnings?: string },\n): Promise<Experiment | null> {\n try {\n const now = new Date().toISOString();\n await sqlite.execute(\n `UPDATE experiments\n SET status = 'completed', end_date = ?, result_value = ?, outcome = ?,\n learnings = ?, updated_at = ?\n WHERE id = ?`,\n [\n now,\n Number(data.result_value) || 0,\n data.outcome,\n data.learnings ?? null,\n now,\n id,\n ],\n );\n const updated = await getExperimentById(id);\n // Closed-loop: a completed experiment propagates to the measure layer\n // (executed-work counter + calendar marker + linked-objective progress refresh).\n if (updated) {\n await propagateExperimentComplete({\n title: updated.title,\n outcome: updated.outcome,\n objective_id: updated.objective_id,\n });\n }\n return updated;\n } catch {\n return null;\n }\n}\n\n// ─── Stats ───────────────────────────────────────────────────────────────────\n\nexport async function getExperimentStats(): Promise<{\n win_rate: number;\n avg_uplift: number;\n experiments_by_stage: Record<FunnelStage, number>;\n top_lever: AARRRLever | null;\n fastest_days: number | null;\n total: number;\n by_status: Record<string, number>;\n}> {\n const fallback = {\n win_rate: 0,\n avg_uplift: 0,\n experiments_by_stage: { TOFU: 0, MOFU: 0, BOFU: 0 } as Record<FunnelStage, number>,\n top_lever: null as AARRRLever | null,\n fastest_days: null as number | null,\n total: 0,\n by_status: {} as Record<string, number>,\n };\n try {\n const all = await listExperiments();\n if (!all.length) return fallback;\n\n const completed = all.filter((e) => e.status === \"completed\");\n const wins = completed.filter((e) => e.outcome === \"win\").length;\n const win_rate = completed.length > 0\n ? Math.round((wins / completed.length) * 100)\n : 0;\n\n const uplifts = completed\n .map((e) => e.uplift_pct)\n .filter((u): u is number => u !== null && Number.isFinite(u));\n const avg_uplift = uplifts.length > 0\n ? Math.round((uplifts.reduce((a, b) => a + b, 0) / uplifts.length) * 100) / 100\n : 0;\n\n const experiments_by_stage: Record<FunnelStage, number> = { TOFU: 0, MOFU: 0, BOFU: 0 };\n for (const e of all) {\n if (e.funnel_stage && experiments_by_stage[e.funnel_stage] != null) {\n experiments_by_stage[e.funnel_stage] += 1;\n }\n }\n\n // Top lever = lever with highest win count among completed experiments.\n const leverWins = new Map<AARRRLever, number>();\n for (const e of completed) {\n if (e.outcome !== \"win\" || !e.aarrr_lever) continue;\n leverWins.set(e.aarrr_lever, (leverWins.get(e.aarrr_lever) || 0) + 1);\n }\n let top_lever: AARRRLever | null = null;\n let topCount = 0;\n for (const [lever, count] of leverWins) {\n if (count > topCount) { top_lever = lever; topCount = count; }\n }\n\n // Fastest completed experiment (start → end in days).\n let fastest_days: number | null = null;\n for (const e of completed) {\n if (!e.start_date || !e.end_date) continue;\n const days = (new Date(e.end_date).getTime() - new Date(e.start_date).getTime()) / 86400000;\n if (days < 0) continue;\n if (fastest_days === null || days < fastest_days) fastest_days = days;\n }\n if (fastest_days !== null) fastest_days = Math.round(fastest_days * 10) / 10;\n\n const by_status: Record<string, number> = {};\n for (const e of all) {\n by_status[e.status] = (by_status[e.status] || 0) + 1;\n }\n\n return {\n win_rate,\n avg_uplift,\n experiments_by_stage,\n top_lever,\n fastest_days,\n total: all.length,\n by_status,\n };\n } catch {\n return fallback;\n }\n}\n\n// ─── LLM-Assisted ICE Scoring ────────────────────────────────────────────────\n\nconst FALLBACK_ICE = {\n impact: 5,\n confidence: 5,\n ease: 5,\n rationale: \"LLM unavailable — defaulted to neutral ICE.\",\n};\n\nfunction parseICEJson(raw: string): {\n impact: number; confidence: number; ease: number; rationale: string;\n} {\n // First try direct JSON parse, then a regex-fenced fallback (handles\n // ```json ... ``` wrapping or extra prose around the object).\n try {\n const parsed = JSON.parse(raw);\n return {\n impact: clamp10(parsed.impact),\n confidence: clamp10(parsed.confidence),\n ease: clamp10(parsed.ease),\n rationale: String(parsed.rationale ?? \"\").slice(0, 500),\n };\n } catch { /* fall through */ }\n const match = raw.match(/\\{[\\s\\S]*\\}/);\n if (match) {\n try {\n const parsed = JSON.parse(match[0]);\n return {\n impact: clamp10(parsed.impact),\n confidence: clamp10(parsed.confidence),\n ease: clamp10(parsed.ease),\n rationale: String(parsed.rationale ?? \"\").slice(0, 500),\n };\n } catch { /* give up */ }\n }\n return { ...FALLBACK_ICE };\n}\n\nexport async function suggestICE(input: {\n title: string;\n hypothesis?: string;\n funnel_stage?: string;\n aarrr_lever?: string;\n objective_id?: string;\n}): Promise<{\n impact: number;\n confidence: number;\n ease: number;\n rationale: string;\n}> {\n try {\n const config = await getLLMConfig();\n if (!config.api_key) return { ...FALLBACK_ICE };\n\n // Pull funnel summary as context so the model has actual numbers, not just\n // theoretical guesses. Trim to keep prompt small.\n let funnelContext = \"\";\n try {\n const summary = await getFunnelSummary();\n const lines: string[] = [];\n lines.push(`Overall funnel health: ${summary.overall_health}/100`);\n if (summary.bottleneck) {\n lines.push(`Bottleneck stage: ${summary.bottleneck.stage} — ${summary.bottleneck.reason}`);\n }\n for (const s of summary.stages) {\n lines.push(`${s.stage} health=${s.health_score} (${s.metrics.length} metrics)`);\n }\n funnelContext = lines.join(\"\\n\");\n } catch { /* funnel unavailable, score on title alone */ }\n\n const system =\n \"You are a growth analyst scoring an experiment idea using the ICE framework. \" +\n \"Return ONLY valid JSON of shape \" +\n '{\"impact\":1-10,\"confidence\":1-10,\"ease\":1-10,\"rationale\":\"<=2 sentences\"}. ' +\n \"No markdown, no commentary outside the JSON.\";\n\n const user =\n `Experiment: ${input.title}\\n` +\n `Hypothesis: ${input.hypothesis ?? \"(none)\"}\\n` +\n `Funnel stage: ${input.funnel_stage ?? \"(unspecified)\"}\\n` +\n `AARRR lever: ${input.aarrr_lever ?? \"(unspecified)\"}\\n\\n` +\n `Current funnel context:\\n${funnelContext || \"(no live funnel data)\"}\\n\\n` +\n `Score this experiment 1-10 on Impact, Confidence, Ease. Return JSON only.`;\n\n const res = await callLLM(\n config,\n [{ role: \"user\", content: user } as any],\n { systemPrompt: system, maxTokens: 800, temperature: 0.3 },\n );\n const parsed = parseICEJson(res.content || \"\");\n return parsed;\n } catch {\n return { ...FALLBACK_ICE };\n }\n}\n",[],[365,837],"ef0da470-5a5f-11f1-a18f-ee650bb23af1","digest.ts","98641374-589e-11f1-bbad-ee650bb23af1",417,["D",1779950350277],"// Weekly growth digest + daily standup generator.\n// digest = LLM-narrated weekly executive brief stored in growth_digests.\n// standup = ephemeral daily delta brief (not persisted).\n//\n// Both fall back gracefully to deterministic-only assembly when LLM times out\n// or is misconfigured, so the UI never blocks on the model.\nimport { sqlite } from \"../db.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nexport function mondayOf(date: Date | string): string {\n const d = typeof date === \"string\" ? new Date(date) : new Date(date.getTime());\n // getDay: 0=Sun..6=Sat. Shift so Monday = 0.\n const dow = (d.getDay() + 6) % 7;\n d.setDate(d.getDate() - dow);\n d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\n// Run a promise with a hard timeout; resolves to null on timeout.\nfunction withTimeout<T>(p: Promise<T>, ms: number): Promise<T | null> {\n return new Promise<T | null>((resolve) => {\n let done = false;\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null); } }, ms);\n p.then(\n (v) => { if (!done) { done = true; clearTimeout(timer); resolve(v); } },\n () => { if (!done) { done = true; clearTimeout(timer); resolve(null); } },\n );\n });\n}\n\n// Safe dynamic-import wrapper. Returns null if the module doesn't exist or\n// throws on load — lets digest still assemble from whatever data IS available.\nasync function safeImport<T = any>(path: string): Promise<T | null> {\n try { return (await import(path)) as T; } catch { return null; }\n}\n\n// Attempt to parse JSON from an LLM reply, tolerating ```json fences and trailing prose.\nfunction tryParseJson(text: string): any | null {\n if (!text) return null;\n let t = text.trim();\n // Strip markdown fences.\n const fence = t.match(/```(?:json)?\\s*([\\s\\S]*?)```/i);\n if (fence) t = fence[1].trim();\n // Find first { ... last } pair.\n const first = t.indexOf(\"{\");\n const last = t.lastIndexOf(\"}\");\n if (first < 0 || last < first) return null;\n const slice = t.slice(first, last + 1);\n try { return JSON.parse(slice); } catch { return null; }\n}\n\n// ─── Deterministic data collection ───────────────────────────────────────────\n\nasync function collectDigestContext() {\n const nsmMod = await safeImport<any>(\"./nsm.ts\");\n const okrsMod = await safeImport<any>(\"./okrs.ts\");\n const expMod = await safeImport<any>(\"./experiments.ts\");\n const intelMod = await safeImport<any>(\"./intelligence.ts\");\n const signalsMod = await safeImport<any>(\"./signals.ts\");\n\n const [nsm, okr, exp, correlations, hot] = await Promise.all([\n nsmMod?.getNSMProgress?.().catch(() => null) ?? Promise.resolve(null),\n okrsMod?.getOKRDashboard?.().catch(() => null) ?? Promise.resolve(null),\n expMod?.getExperimentStats?.().catch(() => null) ?? Promise.resolve(null),\n intelMod?.getCorrelations?.().catch(() => []) ?? Promise.resolve([]),\n signalsMod?.getHotLeads?.().catch(() => []) ?? Promise.resolve([]),\n ]);\n\n return {\n nsm: nsm || null,\n okr: okr || null,\n experiments: exp || null,\n correlations: Array.isArray(correlations) ? correlations.slice(0, 6) : [],\n hot_leads: Array.isArray(hot) ? hot.slice(0, 5) : [],\n };\n}\n\n// Deterministic fallback digest — used when LLM unavailable.\nfunction deterministicDigest(ctx: any, weekOf: string) {\n const nsmValue = ctx.nsm?.current_value ?? null;\n const nsmTarget = ctx.nsm?.target_value ?? null;\n const nsmStatus = (() => {\n if (!ctx.nsm) return \"No North Star metric configured.\";\n if (nsmValue == null || nsmTarget == null) return \"NSM data unavailable this period.\";\n const pct = nsmTarget > 0 ? (nsmValue / nsmTarget) * 100 : 0;\n return `${ctx.nsm.metric_name ?? \"NSM\"} at ${nsmValue} (${pct.toFixed(0)}% of target ${nsmTarget}).`;\n })();\n\n const positives = (ctx.correlations || []).filter((c: any) => c.direction === \"positive\");\n const negatives = (ctx.correlations || []).filter((c: any) => c.direction === \"negative\");\n const topWin = positives[0]?.narrative ?? \"No strong positive correlations detected this week.\";\n const topRisk = negatives[0]?.narrative ?? \"No critical risks flagged.\";\n\n const recommended_experiments: string[] = [];\n if (negatives[0]?.action_prompt) recommended_experiments.push(negatives[0].action_prompt);\n if (positives[0]?.action_prompt) recommended_experiments.push(positives[0].action_prompt);\n if (recommended_experiments.length === 0) {\n recommended_experiments.push(\"Add 1-2 hypothesis-driven experiments to fill the empty backlog.\");\n }\n\n // getOKRDashboard() returns scalar stats ({total,on_track,at_risk,avg_progress,quarter}),\n // not an objectives array — build callouts from those.\n const okr_callouts: string[] = [];\n if (ctx.okr && (ctx.okr.total ?? 0) > 0) {\n okr_callouts.push(`${ctx.okr.total} objective${ctx.okr.total > 1 ? \"s\" : \"\"}${ctx.okr.quarter ? \" (\" + ctx.okr.quarter + \")\" : \"\"}: ${Math.round(ctx.okr.avg_progress ?? 0)}% avg progress.`);\n okr_callouts.push(`${ctx.okr.on_track ?? 0} on track · ${ctx.okr.at_risk ?? 0} at risk.`);\n } else {\n okr_callouts.push(\"No active objectives — set Q-goals to unlock OKR tracking.\");\n }\n\n const next_actions: string[] = [];\n if ((ctx.hot_leads || []).length) {\n next_actions.push(`Follow up with ${ctx.hot_leads.length} hot lead${ctx.hot_leads.length > 1 ? \"s\" : \"\"} (top: ${ctx.hot_leads[0].name || ctx.hot_leads[0].email}).`);\n }\n if (ctx.experiments?.running_count) {\n next_actions.push(`Review ${ctx.experiments.running_count} running experiment${ctx.experiments.running_count > 1 ? \"s\" : \"\"} for early results.`);\n }\n if (next_actions.length === 0) next_actions.push(\"No urgent items — focus on planning next week's bets.\");\n\n const headline = nsmValue != null && nsmTarget != null\n ? `Week of ${weekOf}: NSM at ${nsmValue}/${nsmTarget}, ${ctx.hot_leads?.length || 0} hot leads.`\n : `Week of ${weekOf}: ${ctx.hot_leads?.length || 0} hot leads, ${(ctx.correlations || []).length} signals tracked.`;\n\n return {\n headline,\n nsm_status: nsmStatus,\n top_win: topWin,\n top_risk: topRisk,\n recommended_experiments,\n okr_callouts,\n next_actions,\n };\n}\n\n// ─── Weekly digest ───────────────────────────────────────────────────────────\n\nexport interface GrowthDigest {\n headline: string;\n nsm_status: string;\n top_win: string;\n top_risk: string;\n recommended_experiments: string[];\n okr_callouts: string[];\n next_actions: string[];\n week_of: string;\n generated_at: string;\n}\n\nexport async function generateDigest(): Promise<GrowthDigest> {\n const weekOf = mondayOf(new Date());\n const ctx = await collectDigestContext();\n const fallback = deterministicDigest(ctx, weekOf);\n\n let enriched: any = null;\n try {\n const cfg = await getLLMConfig();\n if (cfg?.api_key) {\n const prompt = [\n \"You are the growth analyst for an early-stage startup. Produce a weekly growth digest as STRICT JSON.\",\n \"Use only the data provided — never invent metrics.\",\n \"Keys (all required): headline (string, <=120 chars), nsm_status (string), top_win (string), top_risk (string),\",\n \"recommended_experiments (array of 2-4 short strings), okr_callouts (array of 1-3 strings), next_actions (array of 2-4 short strings).\",\n \"Respond with ONLY the JSON object, no prose, no markdown fences.\",\n \"\",\n `WEEK_OF: ${weekOf}`,\n `NSM: ${JSON.stringify(ctx.nsm)}`,\n `OKRS: ${JSON.stringify(ctx.okr)}`,\n `EXPERIMENTS: ${JSON.stringify(ctx.experiments)}`,\n `CORRELATIONS: ${JSON.stringify(ctx.correlations)}`,\n `HOT_LEADS: ${JSON.stringify(ctx.hot_leads)}`,\n ].join(\"\\n\");\n\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 1500,\n temperature: 0.4,\n systemPrompt: \"You write tight, executive-level growth briefs. Always respond with valid JSON.\",\n });\n const result = await withTimeout(llmCall, 30000) as { content?: string } | null;\n if (result?.content) {\n enriched = tryParseJson(result.content);\n }\n }\n } catch { /* LLM failed — fallback already prepared */ }\n\n // Merge: enriched fields override fallback ones; missing keys keep fallback.\n const merged = { ...fallback, ...(enriched || {}) };\n\n // Coerce list fields back to arrays in case the LLM returned strings.\n for (const key of [\"recommended_experiments\", \"okr_callouts\", \"next_actions\"] as const) {\n if (!Array.isArray((merged as any)[key])) {\n (merged as any)[key] = (fallback as any)[key];\n }\n }\n\n const generated_at = new Date().toISOString();\n const digest: GrowthDigest = {\n headline: String(merged.headline ?? fallback.headline),\n nsm_status: String(merged.nsm_status ?? fallback.nsm_status),\n top_win: String(merged.top_win ?? fallback.top_win),\n top_risk: String(merged.top_risk ?? fallback.top_risk),\n recommended_experiments: merged.recommended_experiments,\n okr_callouts: merged.okr_callouts,\n next_actions: merged.next_actions,\n week_of: weekOf,\n generated_at,\n };\n\n // Persist (upsert per week_of so re-runs in the same week replace).\n try {\n const nsmValue = ctx.nsm?.current_value ?? null;\n const existing = await sqlite.execute(\"SELECT id FROM growth_digests WHERE week_of = ?\", [weekOf]);\n if (existing.rows?.length) {\n await sqlite.execute(\n \"UPDATE growth_digests SET payload = ?, nsm_value_at_generation = ?, generated_at = ? WHERE week_of = ?\",\n [JSON.stringify(digest), nsmValue, generated_at, weekOf]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO growth_digests (id, week_of, payload, nsm_value_at_generation, generated_at) VALUES (?, ?, ?, ?, ?)\",\n [crypto.randomUUID(), weekOf, JSON.stringify(digest), nsmValue, generated_at]\n );\n }\n } catch { /* non-critical: still return digest */ }\n\n return digest;\n}\n\nexport async function getLatestDigest(): Promise<GrowthDigest | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, week_of, generated_at FROM growth_digests ORDER BY generated_at DESC LIMIT 1\"\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any[];\n let parsed: any = null;\n try { parsed = JSON.parse(String(row[0])); } catch { parsed = null; }\n if (!parsed) return null;\n return {\n headline: parsed.headline ?? \"\",\n nsm_status: parsed.nsm_status ?? \"\",\n top_win: parsed.top_win ?? \"\",\n top_risk: parsed.top_risk ?? \"\",\n recommended_experiments: Array.isArray(parsed.recommended_experiments) ? parsed.recommended_experiments : [],\n okr_callouts: Array.isArray(parsed.okr_callouts) ? parsed.okr_callouts : [],\n next_actions: Array.isArray(parsed.next_actions) ? parsed.next_actions : [],\n week_of: String(row[1]),\n generated_at: String(row[2]),\n };\n } catch { return null; }\n}\n\nexport async function getDigestHistory(limit = 10): Promise<Array<{ week_of: string; headline: string; generated_at: string }>> {\n try {\n const r = await sqlite.execute(\n \"SELECT week_of, payload, generated_at FROM growth_digests ORDER BY week_of DESC LIMIT ?\",\n [limit]\n );\n return (r.rows || []).map((row: any) => {\n let headline = \"\";\n try { headline = String(JSON.parse(String(row[1]))?.headline ?? \"\"); } catch {}\n return { week_of: String(row[0]), headline, generated_at: String(row[2]) };\n });\n } catch { return []; }\n}\n\n// ─── Daily standup (ephemeral) ───────────────────────────────────────────────\n\nexport interface StandupBrief {\n yesterday: string[];\n today: string[];\n blockers: string[];\n signal_alert: string;\n}\n\nasync function collectStandupContext(workspace: string) {\n // tasks moved to done in last 24h\n let doneTasks: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, completed_at FROM tasks\n WHERE workspace = ? AND status = 'done' AND completed_at >= datetime('now','-1 day')\n ORDER BY completed_at DESC LIMIT 20`,\n [workspace]\n );\n doneTasks = (r.rows || []).map((row: any) => ({\n id: row[0], title: row[1], completed_at: row[2],\n }));\n } catch { doneTasks = []; }\n\n // pending tasks for \"today\" plan\n let pendingTasks: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title FROM tasks\n WHERE workspace = ? AND status IN ('todo','doing')\n ORDER BY created_at DESC LIMIT 20`,\n [workspace]\n );\n pendingTasks = (r.rows || []).map((row: any) => ({ id: row[0], title: row[1] }));\n } catch { pendingTasks = []; }\n\n // experiments with status changes in last 24h (table may not exist yet)\n let expChanges: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, status, updated_at FROM experiments\n WHERE updated_at >= datetime('now','-1 day')\n ORDER BY updated_at DESC LIMIT 10`\n );\n expChanges = (r.rows || []).map((row: any) => ({\n id: row[0], name: row[1], status: row[2], updated_at: row[3],\n }));\n } catch { expChanges = []; }\n\n // new intent signals in 24h\n let newSignals: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT lead_email, lead_name, signal_type, signal_strength FROM intent_signals\n WHERE detected_at >= datetime('now','-1 day')\n ORDER BY signal_strength DESC, detected_at DESC LIMIT 10`\n );\n newSignals = (r.rows || []).map((row: any) => ({\n email: row[0], name: row[1], type: row[2], strength: row[3],\n }));\n } catch { newSignals = []; }\n\n // OKR progress changes in 24h (key_results table may not exist yet)\n let krChanges: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, progress, updated_at FROM key_results\n WHERE updated_at >= datetime('now','-1 day')\n ORDER BY updated_at DESC LIMIT 10`\n );\n krChanges = (r.rows || []).map((row: any) => ({\n id: row[0], title: row[1], progress: row[2], updated_at: row[3],\n }));\n } catch { krChanges = []; }\n\n return { doneTasks, pendingTasks, expChanges, newSignals, krChanges };\n}\n\nfunction deterministicStandup(ctx: any): StandupBrief {\n const yesterday: string[] = [];\n for (const t of ctx.doneTasks.slice(0, 5)) yesterday.push(`Shipped: ${t.title}`);\n for (const e of ctx.expChanges.slice(0, 3)) yesterday.push(`Experiment \"${e.name}\" -> ${e.status}`);\n for (const kr of ctx.krChanges.slice(0, 2)) yesterday.push(`KR \"${kr.title}\" now ${Math.round(kr.progress || 0)}%`);\n if (yesterday.length === 0) yesterday.push(\"No recorded activity in the last 24 hours.\");\n\n const today: string[] = [];\n for (const t of ctx.pendingTasks.slice(0, 5)) today.push(t.title);\n if (today.length === 0) today.push(\"Nothing queued — pick the next experiment off the backlog.\");\n\n const blockers: string[] = [];\n if (ctx.pendingTasks.length > 10) blockers.push(`${ctx.pendingTasks.length} open tasks — backlog growing.`);\n if (ctx.expChanges.some((e: any) => e.status === \"blocked\" || e.status === \"failed\")) {\n blockers.push(\"One or more experiments need triage.\");\n }\n\n const strongSignals = (ctx.newSignals || []).filter((s: any) => (s.strength ?? 0) >= 3);\n const signal_alert = strongSignals.length\n ? `${strongSignals.length} high-intent signal${strongSignals.length > 1 ? \"s\" : \"\"} in last 24h — top: ${strongSignals[0].name || strongSignals[0].email} (${strongSignals[0].type}).`\n : ctx.newSignals.length\n ? `${ctx.newSignals.length} new signal${ctx.newSignals.length > 1 ? \"s\" : \"\"}, none high-intent yet.`\n : \"No new intent signals in the last 24 hours.\";\n\n return { yesterday, today, blockers, signal_alert };\n}\n\nexport async function generateStandup(opts: { workspace: string }): Promise<StandupBrief> {\n const workspace = opts.workspace || \"marketing\";\n const ctx = await collectStandupContext(workspace);\n const fallback = deterministicStandup(ctx);\n\n let enriched: any = null;\n try {\n const cfg = await getLLMConfig();\n if (cfg?.api_key) {\n const prompt = [\n \"You are a standup coach. Produce a daily standup brief as STRICT JSON.\",\n \"Keys (all required): yesterday (array of strings), today (array of strings), blockers (array of strings), signal_alert (string).\",\n \"Be concise — each list item under 80 chars. Respond with ONLY the JSON object.\",\n \"\",\n `WORKSPACE: ${workspace}`,\n `DONE_LAST_24H: ${JSON.stringify(ctx.doneTasks)}`,\n `PENDING: ${JSON.stringify(ctx.pendingTasks)}`,\n `EXPERIMENT_CHANGES: ${JSON.stringify(ctx.expChanges)}`,\n `NEW_SIGNALS: ${JSON.stringify(ctx.newSignals)}`,\n `KR_PROGRESS: ${JSON.stringify(ctx.krChanges)}`,\n ].join(\"\\n\");\n\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 700,\n temperature: 0.4,\n systemPrompt: \"You write punchy daily standups. Always respond with valid JSON.\",\n });\n const result = await withTimeout(llmCall, 10000) as { content?: string } | null;\n if (result?.content) enriched = tryParseJson(result.content);\n }\n } catch { /* fallback */ }\n\n if (!enriched) return fallback;\n\n return {\n yesterday: Array.isArray(enriched.yesterday) ? enriched.yesterday : fallback.yesterday,\n today: Array.isArray(enriched.today) ? enriched.today : fallback.today,\n blockers: Array.isArray(enriched.blockers) ? enriched.blockers : fallback.blockers,\n signal_alert: typeof enriched.signal_alert === \"string\" ? enriched.signal_alert : fallback.signal_alert,\n };\n}\n",[],[365,845],"f380acb0-5b58-11f1-8321-ee650bb23af1","deals.ts","6e518b64-5b48-11f1-bbb6-ee650bb23af1",551,["D",1780057302563],"import { sqlite } from \"../db.ts\";\nimport { getLeadScores } from \"./intelligence.ts\";\nimport { recordConversion } from \"./aio.ts\";\n\n// CRM pipeline deals. Hybrid: rows seed from qualified leads (getLeadScores) via syncDealsFromLeads,\n// and are fully user-editable (create/edit/drag stage). See db.ts `deals` table.\n\nexport const DEAL_STAGES = [\"lead\", \"qualified\", \"proposal\", \"negotiation\", \"won\", \"lost\"] as const;\n\nconst STAGE_DEFAULT_PROB: Record<string, number> = {\n lead: 10, qualified: 30, proposal: 50, negotiation: 70, won: 100, lost: 0,\n};\n\n// Lead.stage (cold|MQL|SQL) → pipeline stage. Cold leads are skipped on sync (too noisy for a pipeline).\nconst LEAD_STAGE_MAP: Record<string, string> = { MQL: \"qualified\", SQL: \"proposal\" };\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const o: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) o[columns[i]] = row[i];\n return o;\n}\n\nexport async function listDeals(stage?: string): Promise<any[]> {\n const q = stage\n ? \"SELECT * FROM deals WHERE stage = ? ORDER BY updated_at DESC\"\n : \"SELECT * FROM deals ORDER BY updated_at DESC\";\n const r = await sqlite.execute(q, stage ? [stage] : []);\n return (r.rows || []).map((row) => rowToObj(r.columns, row as any[]));\n}\n\nexport async function getDeal(id: string): Promise<any | null> {\n const r = await sqlite.execute(\"SELECT * FROM deals WHERE id = ?\", [id]);\n if (!r.rows || r.rows.length === 0) return null;\n const deal = rowToObj(r.columns, r.rows[0] as any[]);\n // Activity = this contact's intent signals (the deal's lead history).\n const email = deal.contact_email || deal.lead_email;\n let signals: any[] = [];\n if (email) {\n const s = await sqlite.execute(\n \"SELECT signal_type, signal_source, signal_strength, detected_at FROM intent_signals WHERE lead_email = ? ORDER BY detected_at DESC LIMIT 50\",\n [email],\n ).catch(() => null);\n signals = (s?.rows || []).map((row: any) => rowToObj(s!.columns, row));\n }\n return { ...deal, signals };\n}\n\nexport async function createDeal(data: Record<string, any>): Promise<any> {\n const id = crypto.randomUUID();\n const stage = DEAL_STAGES.includes(data.stage) ? data.stage : \"lead\";\n const now = new Date().toISOString();\n await sqlite.execute(\n `INSERT INTO deals (id, title, company, contact_name, contact_email, stage, amount, currency, probability, owner, close_date, source, lead_email, notes, created_at, updated_at, last_activity_at)\n VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)`,\n [\n id, data.title ?? null, data.company ?? null, data.contact_name ?? null, data.contact_email ?? null,\n stage, Number(data.amount) || 0, data.currency ?? \"USD\",\n data.probability != null ? Number(data.probability) : (STAGE_DEFAULT_PROB[stage] ?? 0),\n data.owner ?? null, data.close_date ?? null, data.source ?? \"manual\",\n data.lead_email ?? data.contact_email ?? null, data.notes ?? null, now, now, now,\n ],\n );\n return getDeal(id);\n}\n\nexport async function updateDeal(id: string, data: Record<string, any>): Promise<any | null> {\n const existing = await getDeal(id);\n if (!existing) return null;\n const stage = data.stage != null && DEAL_STAGES.includes(data.stage) ? data.stage : existing.stage;\n // A bare stage move (drag) auto-bumps probability to the stage default unless probability was sent.\n const prob = data.probability != null\n ? Number(data.probability)\n : (data.stage != null && data.stage !== existing.stage ? (STAGE_DEFAULT_PROB[stage] ?? existing.probability) : existing.probability);\n const merged = {\n title: data.title ?? existing.title,\n company: data.company ?? existing.company,\n contact_name: data.contact_name ?? existing.contact_name,\n contact_email: data.contact_email ?? existing.contact_email,\n stage,\n amount: data.amount != null ? Number(data.amount) : existing.amount,\n currency: data.currency ?? existing.currency,\n probability: prob,\n owner: data.owner ?? existing.owner,\n close_date: data.close_date ?? existing.close_date,\n notes: data.notes ?? existing.notes,\n };\n await sqlite.execute(\n `UPDATE deals SET title=?, company=?, contact_name=?, contact_email=?, stage=?, amount=?, currency=?, probability=?, owner=?, close_date=?, notes=?, updated_at=?, last_activity_at=? WHERE id=?`,\n [merged.title, merged.company, merged.contact_name, merged.contact_email, merged.stage, merged.amount,\n merged.currency, merged.probability, merged.owner, merged.close_date, merged.notes,\n new Date().toISOString(), new Date().toISOString(), id],\n );\n // Revenue interconnection: when a deal is newly won, record its amount as a conversion so it\n // flows into attribution → the funnel \"attribution/revenue\" metric → the revenue OKR Key Result.\n if (merged.stage === \"won\" && existing.stage !== \"won\" && merged.amount > 0) {\n await recordConversion({\n utm_source: merged.source || \"crm\",\n utm_medium: \"deal\",\n utm_campaign: merged.company || merged.title || undefined,\n conversion_value: merged.amount,\n }).catch(() => {});\n }\n // Symmetry: un-winning a deal backs the revenue out so the OKR stays honest.\n if (existing.stage === \"won\" && merged.stage !== \"won\" && existing.amount > 0) {\n await recordConversion({\n utm_source: existing.source || \"crm\",\n utm_medium: \"deal\",\n utm_campaign: existing.company || existing.title || undefined,\n conversion_value: -existing.amount,\n }).catch(() => {});\n }\n return getDeal(id);\n}\n\nexport async function deleteDeal(id: string): Promise<boolean> {\n const r = await sqlite.execute(\"DELETE FROM deals WHERE id = ?\", [id]);\n return Number((r as any)?.rowsAffected ?? 0) > 0;\n}\n\n// Heuristic: is this email a system/notification sender rather than a real prospect? Used to\n// PRE-deselect junk in the interactive import (the user still decides). Imperfect by design.\nexport function looksLikeSystemSender(email: string): boolean {\n const e = String(email || \"\").toLowerCase();\n const local = e.split(\"@\")[0] || \"\";\n const domain = e.split(\"@\")[1] || \"\";\n const localBad = /(^|[._-])(no-?reply|donotreply|do-not-reply|notif(y|ication)?s?|mailer-daemon|postmaster|support|help(desk)?|system|sysadmin|admin|alerts?|info|hello|team|automated|auto|bot|updates?|news(letter)?|billing|account|security|root|webmaster|invites?|calendar)([._-]|$)/.test(local);\n const domainBad = /(microsoft\\.com|sap\\.com|read\\.ai|google\\.com|atlassian\\.|sendgrid|mailchimp|amazonses|salesforce\\.com|zoom\\.us|calendly|docusign|slack\\.com|github\\.com)/.test(domain)\n || /^(mail|email|notif(y|ications?)?|engage|reply|bounce|news|info)\\./.test(domain);\n return localBad || domainBad;\n}\n\nasync function insertDealFromLead(l: any, have: Set<string>, now: string): Promise<boolean> {\n if (!l.email || have.has(l.email)) return false;\n const stage = LEAD_STAGE_MAP[l.stage] || \"lead\";\n await sqlite.execute(\n `INSERT INTO deals (id, title, company, contact_name, contact_email, stage, amount, probability, source, lead_email, last_activity_at, created_at, updated_at)\n VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)`,\n [crypto.randomUUID(), `${l.company || l.name || l.email} — opportunity`, l.company ?? null, l.name ?? null,\n l.email, stage, 0, STAGE_DEFAULT_PROB[stage] ?? 0, l.source || \"lead\", l.email, l.last_touch_at ?? null, now, now],\n );\n have.add(l.email);\n return true;\n}\n\n// Interactive import: list leads not yet in the pipeline, annotated with a prospect guess so the\n// user can review + select (instead of auto-adding notification/system senders).\nexport async function listCandidates(): Promise<any[]> {\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n return (leads as any[]).filter((l) => l.email && !have.has(l.email)).map((l) => {\n const sys = looksLikeSystemSender(l.email);\n return {\n email: l.email, name: l.name, company: l.company, stage: l.stage,\n mql_score: l.mql_score, sql_signals: l.sql_signals || [], last_touch_at: l.last_touch_at,\n likely_prospect: !sys && l.stage !== \"cold\",\n reason: sys ? \"Looks like a system/notification sender\" : (l.stage === \"cold\" ? \"Low engagement (cold)\" : \"Engaged lead\"),\n };\n });\n}\n\n// Create deals only for the explicitly-selected lead emails (the interactive import).\nexport async function importDeals(emails: string[]): Promise<{ created: number }> {\n const want = new Set((emails || []).map(String));\n if (!want.size) return { created: 0 };\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n const now = new Date().toISOString();\n let created = 0;\n for (const l of leads as any[]) {\n if (!want.has(l.email)) continue;\n if (await insertDealFromLead(l, have, now)) created++;\n }\n return { created };\n}\n\n// Auto-seed deals from qualified (MQL/SQL) leads that don't already have a deal. Idempotent.\n// Kept for convenience / chat; the interactive import (listCandidates + importDeals) is the primary path.\nexport async function syncDealsFromLeads(): Promise<{ created: number; leads: number }> {\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n const now = new Date().toISOString();\n let created = 0;\n for (const l of leads as any[]) {\n if (l.stage === \"cold\" || looksLikeSystemSender(l.email)) continue; // skip cold + obvious system senders\n if (await insertDealFromLead(l, have, now)) created++;\n }\n return { created, leads: (leads as any[]).length };\n}\n",[],[365,853],"d435f1b6-5d61-11f1-9b2a-1607ee4eb77e","company.ts","d435f260-5d61-11f1-9b2a-1607ee4eb77e",577,["D",1780281017861],"// Part 15.2 — Company Profile: the app's first-class understanding of the BUSINESS.\n// Until now every prompt described the APP's mechanics, never WHO the company is. This stores\n// a structured profile (positioning, ICP, voice, products, competitors, goals) in app_config\n// under one key, and renders a compact markdown block injected GLOBALLY into every AI surface.\n// It supersedes Part 1 D's flat \"Brand Context\" by giving it structure (with a text alias kept).\nimport { sqlite } from \"../db.ts\";\n\nconst KEY = \"company_profile\";\n\nexport interface CompanyProfile {\n name?: string;\n product?: string; // what it sells (products/services)\n icp?: string; // ideal customer profile, segments, personas\n value_proposition?: string;// positioning / value prop\n competitors?: string;\n voice?: string; // tone/voice + brand do's & don'ts\n differentiators?: string;\n pricing?: string;\n goals?: string; // goals / targets\n markets?: string; // markets / geos\n // free-form fallback (Part 1 D \"Brand Context\" text); used if structured fields are empty.\n notes?: string;\n}\n\nexport async function getCompanyProfile(): Promise<CompanyProfile> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [KEY]);\n const raw = (r.rows?.[0] as any[])?.[0];\n if (!raw) return {};\n const parsed = JSON.parse(String(raw));\n return (parsed && typeof parsed === \"object\") ? parsed : {};\n } catch { return {}; }\n}\n\nexport async function setCompanyProfile(data: CompanyProfile): Promise<CompanyProfile> {\n // Merge with existing so partial PUTs don't wipe other fields.\n const cur = await getCompanyProfile();\n const merged: CompanyProfile = { ...cur };\n for (const [k, v] of Object.entries(data || {})) {\n if (typeof v === \"string\") merged[k as keyof CompanyProfile] = v as any;\n }\n await sqlite.execute(\n `INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, datetime('now'))\n ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = datetime('now')`,\n [KEY, JSON.stringify(merged)]\n );\n return merged;\n}\n\n// Part 1 D alias — Brand Context as plain text round-trips into `notes`.\nexport async function getBrandContext(): Promise<string> {\n const p = await getCompanyProfile();\n return (p.notes ?? \"\").toString();\n}\nexport async function setBrandContext(text: string): Promise<void> {\n await setCompanyProfile({ notes: String(text ?? \"\") });\n}\n\nconst LABELS: Array<[keyof CompanyProfile, string]> = [\n [\"name\", \"Company\"],\n [\"product\", \"What we sell\"],\n [\"value_proposition\", \"Positioning / value prop\"],\n [\"icp\", \"Ideal customer (ICP)\"],\n [\"differentiators\", \"Differentiators\"],\n [\"competitors\", \"Competitors\"],\n [\"pricing\", \"Pricing\"],\n [\"markets\", \"Markets\"],\n [\"goals\", \"Goals / targets\"],\n [\"voice\", \"Voice / tone\"],\n];\n\n// Compact markdown for prompt injection (capped ~2KB per Part 15.4 token budget).\nexport async function getCompanyContextString(maxChars = 2000): Promise<string> {\n const p = await getCompanyProfile();\n const lines: string[] = [];\n for (const [k, label] of LABELS) {\n const v = (p[k] ?? \"\").toString().trim();\n if (v) lines.push(`- ${label}: ${v}`);\n }\n if (p.notes && p.notes.trim() && lines.length === 0) {\n // Pure free-form brand context.\n return (\"## Company Profile\\n\" + p.notes.trim()).slice(0, maxChars);\n }\n if (lines.length === 0) return \"\";\n let out = \"## Company Profile (who this business is — make every output specific to it)\\n\" + lines.join(\"\\n\");\n if (p.notes && p.notes.trim()) out += `\\n- Notes: ${p.notes.trim()}`;\n return out.slice(0, maxChars);\n}\n\nexport function isProfileEmpty(p: CompanyProfile): boolean {\n return !Object.values(p || {}).some((v) => typeof v === \"string\" && v.trim());\n}\n",[],[365,861],"1ef55604-5d69-11f1-b78e-1607ee4eb77e","chat.ts","31ca5a90-56ca-11f1-a61d-ee650bb23af1",610,["D",1780284149737],38,"import { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM, callLLMStream } from \"./llm.ts\";\nimport { getFunnelSummary, getCorrelations } from \"./intelligence.ts\";\nimport { listAgents, createAgent, getAgent } from \"./agents.ts\";\nimport { getCredentials, handleIntegrationProxy } from \"./integrations.ts\";\nimport { enqueueAction } from \"./actions.ts\";\nimport { createTask } from \"./tasks.ts\";\nimport { createCalendarEntry } from \"./calendar_unified.ts\";\nimport { listObjectives, createObjective } from \"./okrs.ts\";\nimport { createExperiment } from \"./experiments.ts\";\nimport type { ChatMessage } from \"../types.ts\";\n\n// ─── Constants ────────────────────────────────────────────────────────────────\n\nconst CORS = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, PUT, DELETE, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n};\n\nconst DEFAULT_SYSTEM_PROMPT =\n \"You are an AI assistant for AIHive Command Center, a marketing and sales platform. \" +\n \"You can analyze data, create agents, and execute tasks across integrations.\";\n\n// Grounds the model in its REAL execution capabilities so it actually uses the tools instead of\n// hallucinating that it is \"blocked\" or \"needs credentials\" (the platform already holds them).\nconst EXECUTION_GUIDE =\n \"EXECUTION CAPABILITIES — IMPORTANT: You can actually DO work via tools. NEVER claim you are \" +\n \"'blocked', that you 'need credentials/site details', or that there is a 'security boundary you cannot bypass' — \" +\n \"the platform already stores all integration credentials. \" +\n \"PUBLIC/IRREVERSIBLE actions QUEUE for one-click human approval in the Approvals inbox — call the matching tool: \" +\n \"prepare_wordpress_post (publish to WordPress), prepare_email (send via Gmail), prepare_outlook_email (send via Outlook), prepare_linkedin_post (post to LinkedIn). \" +\n \"INTERNAL/REVERSIBLE actions run IMMEDIATELY (no approval): create_campaign (start a campaign/project), create_task and schedule_calendar_event (create work), \" +\n \"create_drive_folder / create_google_doc / create_google_sheet (create files in Google Drive/Docs/Sheets). \" +\n \"Use get_integration_data to read live data. None of these require the user to provide credentials. \" +\n \"When the user asks you to publish or send something, call the appropriate prepare_* tool with the FULL content, then tell them it is queued in Approvals. \" +\n \"When the user asks to create a campaign/project or a file, call the matching create_* tool and confirm it was created. \" +\n \"Do NOT fabricate step-by-step 'execution logs', and do NOT ask the user for credentials.\";\n\n// ─── Helpers ──────────────────────────────────────────────────────────────────\n\nfunction json(data: any, status = 200): Response {\n return new Response(JSON.stringify(data), {\n status,\n headers: { \"Content-Type\": \"application/json\", ...CORS },\n });\n}\n\nfunction rowToConversation(\n columns: string[],\n row: any[]\n): { id: string; messages: ChatMessage[]; agent_id?: string; workspace?: string } {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n\n let messages: ChatMessage[] = [];\n if (obj.messages) {\n try {\n messages = JSON.parse(obj.messages);\n } catch {\n messages = [];\n }\n }\n\n return {\n id: obj.id,\n messages,\n agent_id: obj.agent_id ?? undefined,\n workspace: obj.workspace ?? undefined,\n };\n}\n\n// ─── Conversation DB ops ──────────────────────────────────────────────────────\n\nasync function loadConversation(\n conversationId: string\n): Promise<{ id: string; messages: ChatMessage[]; agent_id?: string; workspace?: string } | null> {\n const result = await sqlite.execute(\n \"SELECT id, messages, agent_id, workspace FROM conversations WHERE id = ?\",\n [conversationId]\n );\n\n if (!result.rows || result.rows.length === 0) {\n return null;\n }\n\n return rowToConversation(result.columns, result.rows[0]);\n}\n\nasync function saveConversation(\n id: string,\n messages: ChatMessage[],\n agentId?: string,\n workspace?: string\n): Promise<void> {\n const now = new Date().toISOString();\n const messagesJson = JSON.stringify(messages);\n\n const existing = await sqlite.execute(\n \"SELECT id FROM conversations WHERE id = ?\",\n [id]\n );\n\n const ws = workspace ?? \"general\";\n if (existing.rows && existing.rows.length > 0) {\n await sqlite.execute(\n \"UPDATE conversations SET messages = ?, agent_id = ?, workspace = ?, updated_at = ? WHERE id = ?\",\n [messagesJson, agentId ?? null, ws, now, id]\n );\n } else {\n await sqlite.execute(\n `INSERT INTO conversations (id, messages, agent_id, workspace, created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?)`,\n [id, messagesJson, agentId ?? null, ws, now, now]\n );\n }\n}\n\n// ─── Tool definitions ─────────────────────────────────────────────────────────\n\nexport function buildTools(): any[] {\n return [\n {\n type: \"function\",\n function: {\n name: \"get_integration_data\",\n description:\n \"Fetch data from ANY connected integration. Marketing/SEO: ahrefs, ga4, gsc, youtube, wordpress, producthunt, linkedin. \" +\n \"Microsoft 365 (all share one connected account): outlook (inbox/calendar/contacts), onedrive (files), \" +\n \"teams (teams/channels/messages/chats), excel (workbooks/worksheets/range). Google Workspace: gmail, drive, sheets, docs, calendar.\",\n parameters: {\n type: \"object\",\n properties: {\n name: {\n type: \"string\",\n description:\n \"Integration name: ahrefs, ga4, gsc, linkedin, outlook, onedrive, teams, excel, wordpress, youtube, producthunt, gmail, drive, sheets, docs, calendar\",\n },\n endpoint: {\n type: \"string\",\n description:\n \"The endpoint within the integration, e.g. onedrive 'recent'/'files'/'search'; teams 'teams'/'channels'/'messages'/'chats'; excel 'workbooks'/'worksheets'/'range'; outlook 'inbox'/'calendar'/'contacts'; ahrefs 'site-explorer'/'keywords'; gsc 'search-analytics'; linkedin 'posts'.\",\n },\n params: {\n type: \"object\",\n description: \"Additional query parameters to forward to the API\",\n additionalProperties: { type: \"string\" },\n },\n },\n required: [\"name\", \"endpoint\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_agent\",\n description:\n \"Create a new AI agent in AIHive Command Center with a specific role, workspace, and toolset.\",\n parameters: {\n type: \"object\",\n properties: {\n name: {\n type: \"string\",\n description: \"Name for the new agent\",\n },\n workspace: {\n type: \"string\",\n enum: [\"marketing\", \"sales\"],\n description: \"The workspace this agent belongs to\",\n },\n description: {\n type: \"string\",\n description: \"Short description of what this agent does\",\n },\n system_prompt: {\n type: \"string\",\n description: \"System prompt that defines the agent's behavior\",\n },\n tools: {\n type: \"array\",\n items: { type: \"string\" },\n description:\n \"List of tool names the agent has access to (e.g. ['get_integration_data', 'compose_content'])\",\n },\n },\n required: [\"name\", \"workspace\", \"description\", \"system_prompt\", \"tools\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"list_agents\",\n description: \"List all AI agents, optionally filtered by workspace.\",\n parameters: {\n type: \"object\",\n properties: {\n workspace: {\n type: \"string\",\n enum: [\"marketing\", \"sales\"],\n description: \"Filter agents by workspace (optional)\",\n },\n },\n required: [],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"list_objectives\",\n description: \"List OKR objectives (id + title). Use before create_campaign to link a campaign to an existing objective.\",\n parameters: {\n type: \"object\",\n properties: {\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"], description: \"Filter by workspace (optional)\" },\n },\n required: [],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"compose_content\",\n description:\n \"Generate marketing or sales content such as blog posts, LinkedIn updates, emails, or YouTube descriptions.\",\n parameters: {\n type: \"object\",\n properties: {\n type: {\n type: \"string\",\n enum: [\"blog\", \"linkedin\", \"email\", \"youtube_description\"],\n description: \"The type of content to generate\",\n },\n topic: {\n type: \"string\",\n description: \"The topic or subject of the content\",\n },\n tone: {\n type: \"string\",\n description:\n \"The tone of voice for the content (e.g. 'professional', 'casual', 'persuasive')\",\n },\n },\n required: [\"type\", \"topic\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_email\",\n description:\n \"Draft an email and QUEUE it for human approval (Gmail). It is NOT sent until the user approves it in Approvals. Use this to actually act on email outreach.\",\n parameters: {\n type: \"object\",\n properties: {\n to: { type: \"string\", description: \"Recipient email address\" },\n subject: { type: \"string\", description: \"Email subject\" },\n body: { type: \"string\", description: \"Email body (plain text)\" },\n cc: { type: \"string\", description: \"Optional CC address\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once sent\" },\n },\n required: [\"to\", \"subject\", \"body\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_linkedin_post\",\n description:\n \"Draft a LinkedIn post and QUEUE it for human approval. NOT posted until the user approves.\",\n parameters: {\n type: \"object\",\n properties: {\n text: { type: \"string\", description: \"The post text\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once posted\" },\n },\n required: [\"text\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_wordpress_post\",\n description:\n \"Draft a WordPress blog post and QUEUE it for human approval. NOT published until the user approves.\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\", description: \"Post title\" },\n content: { type: \"string\", description: \"Post HTML/markdown content\" },\n status: { type: \"string\", enum: [\"publish\", \"draft\"], description: \"publish (default) or draft\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once published\" },\n },\n required: [\"title\", \"content\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_task\",\n description:\n \"Create a task in the task board. Runs immediately (internal, no approval). Optionally link it to an objective, key result, or experiment.\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\" },\n description: { type: \"string\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n priority: { type: \"string\", enum: [\"low\", \"medium\", \"high\", \"urgent\"] },\n due_date: { type: \"string\", description: \"YYYY-MM-DD\" },\n objective_id: { type: \"string\" },\n key_result_id: { type: \"string\" },\n experiment_id: { type: \"string\" },\n },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"schedule_calendar_event\",\n description:\n \"Add an event to the unified calendar. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\" },\n event_date: { type: \"string\", description: \"YYYY-MM-DD\" },\n event_time: { type: \"string\", description: \"HH:MM (optional)\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n },\n required: [\"title\", \"event_date\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_campaign\",\n description:\n \"Create a marketing/sales campaign (a campaign IS an agent). Runs immediately (internal, no approval). \" +\n \"A campaign is interconnected: it links to an OKR objective and gets an ICE experiment so its impact is measured. \" +\n \"Pass objective_id to link an existing objective (use list_objectives first), or omit to auto-create one from objective_title. \" +\n \"Always provide impact/confidence/ease (1-10) so the linked ICE experiment is scored.\",\n parameters: {\n type: \"object\",\n properties: {\n name: { type: \"string\", description: \"Campaign name\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n channel: { type: \"string\", description: \"linkedin | email | content | blog | seo | social | paid | mixed\" },\n goal: { type: \"string\", description: \"What the campaign aims to achieve\" },\n audience: { type: \"string\", description: \"Target audience (optional)\" },\n status: { type: \"string\", enum: [\"active\", \"paused\", \"draft\", \"completed\"], description: \"default active\" },\n objective_id: { type: \"string\", description: \"Existing OKR objective id to link (optional)\" },\n objective_title: { type: \"string\", description: \"If no objective_id, auto-create an objective with this title (optional)\" },\n quarter: { type: \"string\", description: \"Quarter for the auto-created objective, e.g. 'Q3 2025' (optional; defaults to current quarter)\" },\n funnel_stage: { type: \"string\", enum: [\"TOFU\", \"MOFU\", \"BOFU\"], description: \"Funnel stage for the ICE experiment\" },\n impact: { type: \"number\", description: \"ICE impact 1-10\" },\n confidence: { type: \"number\", description: \"ICE confidence 1-10\" },\n ease: { type: \"number\", description: \"ICE ease 1-10\" },\n },\n required: [\"name\", \"channel\", \"goal\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_drive_folder\",\n description: \"Create a folder in Google Drive. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: { name: { type: \"string\", description: \"Folder name\" } },\n required: [\"name\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_google_doc\",\n description: \"Create a Google Doc, optionally with initial content. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\", description: \"Document title\" },\n content: { type: \"string\", description: \"Optional initial body text\" },\n },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_google_sheet\",\n description: \"Create a Google Sheet. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: { title: { type: \"string\", description: \"Spreadsheet title\" } },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_outlook_email\",\n description:\n \"Draft an email and QUEUE it for human approval (Outlook/Microsoft 365). NOT sent until the user approves it in Approvals.\",\n parameters: {\n type: \"object\",\n properties: {\n to: { type: \"string\", description: \"Recipient email address\" },\n subject: { type: \"string\", description: \"Email subject\" },\n body: { type: \"string\", description: \"Email body (plain text)\" },\n cc: { type: \"string\", description: \"Optional CC address\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once sent\" },\n },\n required: [\"to\", \"subject\", \"body\"],\n },\n },\n },\n ];\n}\n\n// ─── Tool executor ────────────────────────────────────────────────────────────\n\n// Run a WRITE/mutate endpoint on an integration immediately, POSTing through the same proxy\n// the approval queue uses. Returns parsed JSON (or {error}).\nasync function runIntegrationWrite(integration: string, endpoint: string, payload: any): Promise<any> {\n const creds = await getCredentials(integration);\n if (!creds) return { error: `Integration '${integration}' is not connected. Add it in Settings first.` };\n const params = new URLSearchParams({ endpoint });\n const req = new Request(\n `https://aihive.internal/api/integrations/${integration}/data?endpoint=${encodeURIComponent(endpoint)}`,\n { method: \"POST\", headers: { \"Content-Type\": \"application/json\" }, body: JSON.stringify(payload ?? {}) },\n );\n try {\n const res = await handleIntegrationProxy(req, integration, params, creds);\n return await res.json().catch(() => ({}));\n } catch (err: any) {\n return { error: err?.message ?? \"Integration call failed\" };\n }\n}\n\nexport async function executeTool(\n toolName: string,\n toolArgs: Record<string, any>,\n llmConfig: Awaited<ReturnType<typeof getLLMConfig>>\n): Promise<string> {\n switch (toolName) {\n case \"get_integration_data\": {\n const { name, endpoint, params = {} } = toolArgs as {\n name: string;\n endpoint: string;\n params?: Record<string, string>;\n };\n\n const creds = await getCredentials(name);\n if (!creds) {\n return JSON.stringify({\n error: `Integration '${name}' is not configured. Please add credentials first.`,\n });\n }\n\n // Build a synthetic URLSearchParams from args\n const searchParams = new URLSearchParams({ endpoint });\n for (const [k, v] of Object.entries(params)) {\n searchParams.set(k, String(v));\n }\n\n // Build a minimal Request so handlers can read body if needed\n const syntheticReq = new Request(\"https://aihive.internal/proxy\", {\n method: \"GET\",\n });\n\n try {\n const proxyRes = await handleIntegrationProxy(\n syntheticReq,\n name,\n searchParams,\n creds\n );\n const data = await proxyRes.json();\n return JSON.stringify(data);\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Proxy request failed\" });\n }\n }\n\n case \"create_agent\": {\n const { name, workspace, description, system_prompt, tools } = toolArgs as {\n name: string;\n workspace: \"marketing\" | \"sales\";\n description: string;\n system_prompt: string;\n tools: string[];\n };\n\n try {\n const agent = await createAgent({\n name,\n workspace,\n description,\n system_prompt,\n tools,\n });\n return JSON.stringify({ success: true, agent });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create agent\" });\n }\n }\n\n case \"list_agents\": {\n const { workspace } = toolArgs as { workspace?: string };\n try {\n const agents = await listAgents(workspace);\n return JSON.stringify({ agents });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to list agents\" });\n }\n }\n\n case \"compose_content\": {\n const { type, topic, tone = \"professional\" } = toolArgs as {\n type: \"blog\" | \"linkedin\" | \"email\" | \"youtube_description\";\n topic: string;\n tone?: string;\n };\n\n const contentPrompts: Record<string, string> = {\n blog: `Write a comprehensive, SEO-friendly blog post about: \"${topic}\". Tone: ${tone}. Include an introduction, 3-5 sections with headers, and a conclusion.`,\n linkedin: `Write an engaging LinkedIn post about: \"${topic}\". Tone: ${tone}. Keep it under 300 words, include relevant hashtags, and a call-to-action.`,\n email: `Write a professional email about: \"${topic}\". Tone: ${tone}. Include subject line, greeting, body paragraphs, and a clear call-to-action.`,\n youtube_description: `Write an optimized YouTube video description for a video about: \"${topic}\". Tone: ${tone}. Include a hook, timestamps placeholder, links section, and relevant tags.`,\n };\n\n const prompt = contentPrompts[type] ?? `Write content about: \"${topic}\". Tone: ${tone}.`;\n\n try {\n const contentMessages: ChatMessage[] = [\n { role: \"user\", content: prompt },\n ];\n\n const result = await callLLM(llmConfig, contentMessages, {\n systemPrompt:\n \"You are an expert content writer for marketing and sales teams. Produce high-quality, engaging content tailored to the specified format and tone.\",\n maxTokens: 2048,\n });\n\n return JSON.stringify({\n type,\n topic,\n tone,\n content: result.content,\n });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to compose content\" });\n }\n }\n\n case \"prepare_email\": {\n const { to, subject, body, cc, task_id } = toolArgs as any;\n if (!to || !subject) return JSON.stringify({ error: \"to + subject required\" });\n const action = await enqueueAction({\n integration: \"gmail\", endpoint: \"send\",\n payload: { to, subject, body: body ?? \"\", cc },\n title: `Email: ${subject}`,\n preview: `To: ${to}\\nSubject: ${subject}\\n\\n${String(body ?? \"\").slice(0, 600)}`,\n task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"Email drafted and QUEUED FOR APPROVAL — it will NOT send until the user approves it in Approvals.\" });\n }\n\n case \"prepare_linkedin_post\": {\n const { text, task_id } = toolArgs as any;\n if (!text) return JSON.stringify({ error: \"text required\" });\n const action = await enqueueAction({\n integration: \"linkedin\", endpoint: \"create-post\",\n payload: { text },\n title: \"LinkedIn post\",\n preview: String(text).slice(0, 600),\n task_id, source_type: \"chat\", workspace: \"sales\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"LinkedIn post drafted and QUEUED FOR APPROVAL.\" });\n }\n\n case \"prepare_wordpress_post\": {\n const { title, content, status, task_id } = toolArgs as any;\n if (!title || !content) return JSON.stringify({ error: \"title + content required\" });\n const action = await enqueueAction({\n integration: \"wordpress\", endpoint: \"publish-post\",\n payload: { title, content, status: status === \"draft\" ? \"draft\" : \"publish\" },\n title: `WordPress: ${title}`,\n preview: `${title}\\n\\n${String(content).slice(0, 600)}`,\n task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"WordPress post drafted and QUEUED FOR APPROVAL.\" });\n }\n\n case \"create_task\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n try {\n const task = await createTask({\n title: a.title, description: a.description ?? null,\n priority: a.priority, workspace: a.workspace ?? \"marketing\",\n due_date: a.due_date, objective_id: a.objective_id,\n key_result_id: a.key_result_id, experiment_id: a.experiment_id,\n } as any);\n return JSON.stringify({ success: true, task });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create task\" });\n }\n }\n\n case \"schedule_calendar_event\": {\n const a = toolArgs as any;\n if (!a.title || !a.event_date) return JSON.stringify({ error: \"title + event_date (YYYY-MM-DD) required\" });\n try {\n const entry = await createCalendarEntry({\n title: a.title, event_date: a.event_date, event_time: a.event_time ?? null,\n workspace: a.workspace ?? \"marketing\", event_type: \"custom\",\n } as any);\n return JSON.stringify({ success: true, entry });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to schedule event\" });\n }\n }\n\n case \"list_objectives\": {\n const { workspace } = toolArgs as { workspace?: string };\n try {\n const objectives = await listObjectives(workspace ? { workspace } : {});\n return JSON.stringify({ objectives: objectives.map((o: any) => ({ id: o.id, title: o.title, status: o.status, progress: o.progress })) });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to list objectives\" });\n }\n }\n\n case \"create_campaign\": {\n const a = toolArgs as any;\n if (!a.name) return JSON.stringify({ error: \"name required\" });\n const workspace = a.workspace === \"sales\" ? \"sales\" : \"marketing\";\n try {\n // 1. OKR linkage — link an existing objective, else auto-create one (a campaign MUST tie to an OKR).\n let objectiveId = a.objective_id ? String(a.objective_id) : null;\n if (!objectiveId) {\n const now = new Date();\n const quarter = a.quarter || `Q${Math.floor(now.getMonth() / 3) + 1} ${now.getFullYear()}`;\n const obj = await createObjective({\n title: a.objective_title || `${a.name} objective`,\n description: a.goal ?? null,\n quarter,\n workspace,\n status: \"active\",\n } as any);\n objectiveId = obj.id;\n }\n // 2. ICE experiment — connects the campaign's impact to the objective so it gets measured.\n const experiment = await createExperiment({\n title: `${a.name} (campaign)`,\n hypothesis: a.goal ?? null,\n objective_id: objectiveId,\n funnel_stage: a.funnel_stage ?? null,\n channel: a.channel ?? null,\n status: \"idea\",\n impact: a.impact, confidence: a.confidence, ease: a.ease,\n } as any);\n // 3. The campaign itself (a campaign IS an agent).\n const campaign = await createAgent({\n name: a.name, workspace,\n description: a.goal ?? a.description ?? null,\n channel: a.channel ?? \"mixed\",\n goal: a.goal ?? null,\n audience: a.audience ?? null,\n status: a.status ?? \"active\",\n } as any);\n return JSON.stringify({\n success: true,\n campaign,\n linked: { objective_id: objectiveId, experiment_id: experiment.id },\n note: \"Campaign created and connected to an OKR objective + ICE experiment.\",\n });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create campaign\" });\n }\n }\n\n case \"create_drive_folder\": {\n const a = toolArgs as any;\n if (!a.name) return JSON.stringify({ error: \"name required\" });\n const data = await runIntegrationWrite(\"drive\", \"createFolder\", { name: a.name });\n return JSON.stringify(data?.error ? data : { success: true, file: data });\n }\n\n case \"create_google_doc\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n const data = await runIntegrationWrite(\"docs\", \"create\", { title: a.title, content: a.content });\n return JSON.stringify(data?.error ? data : { success: true, doc: data });\n }\n\n case \"create_google_sheet\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n const data = await runIntegrationWrite(\"sheets\", \"create\", { title: a.title });\n return JSON.stringify(data?.error ? data : { success: true, sheet: data });\n }\n\n case \"prepare_outlook_email\": {\n const a = toolArgs as any;\n if (!a.to || !a.subject) return JSON.stringify({ error: \"to + subject required\" });\n const action = await enqueueAction({\n integration: \"outlook\", endpoint: \"send\",\n payload: {\n message: {\n subject: a.subject,\n body: { contentType: \"Text\", content: a.body ?? \"\" },\n toRecipients: [{ emailAddress: { address: a.to } }],\n ...(a.cc ? { ccRecipients: [{ emailAddress: { address: a.cc } }] } : {}),\n },\n saveToSentItems: true,\n },\n title: `Outlook email: ${a.subject}`,\n preview: `To: ${a.to}\\nSubject: ${a.subject}\\n\\n${String(a.body ?? \"\").slice(0, 600)}`,\n task_id: a.task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"Outlook email drafted and QUEUED FOR APPROVAL.\" });\n }\n\n default:\n return JSON.stringify({ error: `Unknown tool: ${toolName}` });\n }\n}\n\n// ─── Main chat handler ────────────────────────────────────────────────────────\n\nexport async function handleChat(req: Request): Promise<Response> {\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS });\n }\n\n if (req.method !== \"POST\") {\n return json({ error: \"Method not allowed\" }, 405);\n }\n\n let body: any;\n try {\n body = await req.json();\n } catch {\n return json({ error: \"Invalid JSON body\" }, 400);\n }\n\n const {\n message,\n agent_id,\n conversation_id: incomingConversationId,\n workspace,\n } = body as {\n message?: string;\n agent_id?: string;\n conversation_id?: string;\n workspace?: string;\n };\n\n if (!message || typeof message !== \"string\" || message.trim() === \"\") {\n return json({ error: \"message field is required and must be a non-empty string\" }, 400);\n }\n\n // Load LLM config\n const llmConfig = await getLLMConfig();\n\n // Load agent if specified\n let agent = null;\n if (agent_id) {\n agent = await getAgent(agent_id);\n }\n\n // Determine conversation ID\n const conversationId = incomingConversationId ?? crypto.randomUUID();\n\n // Load existing conversation history or start fresh\n let conversationHistory: ChatMessage[] = [];\n if (incomingConversationId) {\n const existing = await loadConversation(incomingConversationId);\n if (existing) {\n conversationHistory = existing.messages;\n }\n }\n\n // Add the new user message\n conversationHistory.push({ role: \"user\", content: message.trim() });\n\n // Build funnel-aware system prompt: inject live funnel state + top correlations\n // so the LLM always answers in causal/positional terms (TOFU/MOFU/BOFU).\n // Falls back silently if intelligence layer is unavailable.\n let funnelContext = \"\";\n try {\n const [summary, correlations] = await Promise.all([getFunnelSummary(), getCorrelations()]);\n const stageLines = summary.stages.map((s) =>\n ` ${s.stage} (${s.label}): health ${s.health_score}/100 · ${s.metrics.length} metrics · ${s.diagnosis}`\n ).join(\"\\n\");\n const bottleneck = summary.bottleneck ? `Bottleneck: ${summary.bottleneck.stage} — ${summary.bottleneck.reason}` : \"No bottleneck detected.\";\n const corrLines = correlations.slice(0, 5).map((c) =>\n ` - ${c.narrative} [${c.funnel_stage_from}→${c.funnel_stage_to}, ${c.strength}, lag ${c.lag_days}d, conf ${Math.round(c.confidence*100)}%]`\n ).join(\"\\n\");\n funnelContext = `\n\n## Current Funnel State (live, last 30d)\n${stageLines}\n${bottleneck}\nIntegrations connected: ${summary.total_integrations_connected} · Metrics observed: ${summary.total_metrics_observed}\n\n## Active Signal Correlations\n${corrLines || \" (no correlations yet — connect more integrations)\"}\n\n## Your Role\nYou are **Nexus** — a senior marketing & sales intelligence analyst embedded in the dashboard.\n- When a metric is mentioned, explain its funnel position (TOFU/MOFU/BOFU) and causal relationships\n- Proactively surface the highest-priority bottleneck in every response\n- Suggest specific actions tied to the integration data above\n- Use academic terminology: CAC, LTV, MQL, SQL, attribution, conversion velocity, message-market fit\n- End responses with a concrete next action when applicable\n`.trim();\n } catch { /* intelligence layer unavailable — proceed without context */ }\n\n // I3: agent role prompts. When agent.role is one of listener/topic/creator/analyst, prepend\n // the specialized prefix so the model behaves consistently with the role badge shown in the UI.\n const AGENT_ROLE_PROMPTS: Record<string, string> = {\n listener: `You are a Listener Agent. Your job:\n1. Monitor every incoming signal from Gmail, Outlook, and LinkedIn for buyer intent.\n2. Flag mentions of: pain points, competitor references, pricing questions, budget signals.\n3. Surface weekly patterns (\"3 leads this week mentioned X\").\n4. Err on the side of flagging — never miss a commercially-relevant signal.`,\n topic: `You are a Topic Agent. Your job:\n1. Take buyer pain points + intent signals (from the Listener Agent or Signals workspace).\n2. Generate content ideas: blog titles, LinkedIn angles, email subject lines.\n3. Score each idea by search-volume potential (Ahrefs), funnel-stage fit, estimated ICE impact.\n4. Output a prioritized content backlog formatted as: Title | Funnel Stage | AARRR Lever | Est. ICE Score.\n5. Link every output to an active OKR Key Result when possible.`,\n creator: `You are a Creator Agent. Your job:\n1. Take approved topic ideas and draft complete marketing assets.\n2. Match brand voice: direct, data-backed, B2B SaaS tone.\n3. Blog posts: H1/H2 structure, 800–1200 words, CTA at end.\n4. LinkedIn posts: hook in first line, insight in body, question as CTA, max 250 words.\n5. Email sequences: subject + 3-email sequence (awareness → consideration → CTA).\n6. Always cite one data point from the connected integrations as social proof.`,\n analyst: `You are an Analyst Agent. Your job:\n1. Analyze the funnel summary across all 14 integrations.\n2. Identify the single biggest bottleneck in the current TOFU/MOFU/BOFU funnel.\n3. Propose the top 3 ICE-scored experiments to address it.\n4. Track experiment outcomes and update affected OKR Key Results.\n5. Produce a weekly growth digest: wins, losses, signals, recommended experiments.`,\n };\n const rolePrompt = agent?.role && AGENT_ROLE_PROMPTS[agent.role] ? AGENT_ROLE_PROMPTS[agent.role] : \"\";\n\n const baseSystemPrompt: string =\n agent?.system_prompt?.trim()\n ? agent.system_prompt.trim()\n : DEFAULT_SYSTEM_PROMPT;\n const systemPrompt = [rolePrompt, baseSystemPrompt, EXECUTION_GUIDE, funnelContext]\n .filter((s) => s && s.length > 0)\n .join(\"\\n\\n\");\n\n // Determine tools — always give all tools; agents narrow by integration name via get_integration_data\n // Agent.tools stores integration names (e.g. [\"ga4\",\"gsc\"]), not tool function names,\n // so filtering by function name would always produce an empty list. Give all tools instead.\n const tools = buildTools();\n\n // Read temperature + frequency_penalty from app_config so Settings sliders are real, not cosmetic.\n let temperature: number | undefined = undefined;\n let frequencyPenalty: number | undefined = undefined;\n try {\n const cfgRows = await sqlite.execute(\"SELECT key, value FROM app_config WHERE key IN ('temperature','frequency_penalty')\");\n for (const row of cfgRows.rows) {\n const [k, v] = row as any[];\n const num = parseFloat(String(v).replace(/^\"|\"$/g, \"\"));\n if (!isNaN(num)) {\n if (k === \"temperature\") temperature = num;\n if (k === \"frequency_penalty\") frequencyPenalty = num;\n }\n }\n } catch { /* fall back to provider defaults */ }\n\n // First LLM call\n let llmResult = await callLLM(llmConfig, conversationHistory, {\n systemPrompt,\n tools,\n maxTokens: 4096,\n temperature,\n frequencyPenalty,\n });\n\n const allToolCalls: any[] = [];\n\n // Handle tool use — bounded multi-round loop (was a single round, which truncated\n // any plan needing chained tool calls). Cap rounds to avoid runaway cost/latency.\n let toolRounds = 0;\n const MAX_TOOL_ROUNDS = 5;\n while (llmResult.tool_calls && llmResult.tool_calls.length > 0 && toolRounds < MAX_TOOL_ROUNDS) {\n toolRounds++;\n // Record the assistant message with tool calls.\n // reasoning_content must be echoed back for DeepSeek thinking-mode multi-turn.\n const assistantMsg: any = {\n role: \"assistant\",\n content: llmResult.content || \"\",\n tool_calls: llmResult.tool_calls,\n };\n if ((llmResult as any).reasoning_content) {\n assistantMsg.reasoning_content = (llmResult as any).reasoning_content;\n }\n conversationHistory.push(assistantMsg);\n\n // Execute each tool call and collect results\n const toolResults: Array<{ tool_call_id: string; role: \"tool\"; content: string }> = [];\n\n for (const toolCall of llmResult.tool_calls) {\n allToolCalls.push(toolCall);\n\n let args: Record<string, any> = {};\n try {\n args =\n typeof toolCall.function.arguments === \"string\"\n ? JSON.parse(toolCall.function.arguments)\n : toolCall.function.arguments ?? {};\n } catch {\n args = {};\n }\n\n const toolResult = await executeTool(toolCall.function.name, args, llmConfig);\n\n toolResults.push({\n tool_call_id: toolCall.id,\n role: \"tool\",\n content: toolResult,\n });\n }\n\n // Add tool results to history\n for (const tr of toolResults) {\n conversationHistory.push(tr as any);\n }\n\n // Second LLM call with tool results\n llmResult = await callLLM(llmConfig, conversationHistory, {\n systemPrompt,\n tools,\n maxTokens: 4096,\n temperature,\n frequencyPenalty,\n });\n }\n\n // Add final assistant response to history\n conversationHistory.push({\n role: \"assistant\",\n content: llmResult.content,\n });\n\n // Persist conversation\n await saveConversation(\n conversationId,\n conversationHistory,\n agent_id,\n workspace\n );\n\n return json({\n response: llmResult.content,\n conversation_id: conversationId,\n tool_calls: allToolCalls.length > 0 ? allToolCalls : undefined,\n agent_id: agent_id ?? undefined,\n });\n}\n\n// ─── Streaming chat handler (SSE) ────────────────────────────────────────────\n\nexport async function handleChatStream(req: Request): Promise<Response> {\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS });\n }\n if (req.method !== \"POST\") {\n return new Response(JSON.stringify({ error: \"Method not allowed\" }), { status: 405, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n let body: any;\n try { body = await req.json(); } catch {\n return new Response(JSON.stringify({ error: \"Invalid JSON body\" }), { status: 400, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n const { message, agent_id, conversation_id: incomingConversationId, workspace } = body as {\n message?: string; agent_id?: string; conversation_id?: string; workspace?: string;\n };\n\n if (!message || typeof message !== \"string\" || message.trim() === \"\") {\n return new Response(JSON.stringify({ error: \"message field is required\" }), { status: 400, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n const llmConfig = await getLLMConfig();\n let agent = null;\n if (agent_id) { agent = await getAgent(agent_id); }\n\n const conversationId = incomingConversationId ?? crypto.randomUUID();\n let conversationHistory: ChatMessage[] = [];\n if (incomingConversationId) {\n const existing = await loadConversation(incomingConversationId);\n if (existing) conversationHistory = existing.messages;\n }\n conversationHistory.push({ role: \"user\", content: message.trim() });\n\n const baseSystemPrompt: string = agent?.system_prompt?.trim() ? agent.system_prompt.trim() :\n \"You are an AI assistant for AIHive Command Center, a marketing and sales platform.\";\n\n const encoder = new TextEncoder();\n const stream = new ReadableStream({\n async start(controller) {\n const send = (data: object) => {\n controller.enqueue(encoder.encode(`data: ${JSON.stringify(data)}\\n\\n`));\n };\n\n let fullResponse = \"\";\n try {\n const gen = callLLMStream(llmConfig, conversationHistory, {\n systemPrompt: baseSystemPrompt, maxTokens: 4096,\n });\n for await (const token of gen) {\n if (token) {\n fullResponse += token;\n send({ delta: token, done: false });\n }\n }\n // Persist the full conversation\n conversationHistory.push({ role: \"assistant\", content: fullResponse });\n await saveConversation(conversationId, conversationHistory, agent_id, workspace);\n send({ delta: \"\", done: true, response: fullResponse, conversation_id: conversationId });\n } catch (err: any) {\n send({ delta: \"\", done: true, error: err?.message ?? \"Stream error\", conversation_id: conversationId });\n } finally {\n controller.close();\n }\n },\n });\n\n return new Response(stream, {\n headers: {\n \"Content-Type\": \"text/event-stream\",\n \"Cache-Control\": \"no-cache\",\n \"Connection\": \"keep-alive\",\n ...CORS,\n },\n });\n}\n\n// ─── Conversation history CRUD ────────────────────────────────────────────────\n\nexport async function listConversations(): Promise<any[]> {\n try {\n const result = await sqlite.execute(\n `SELECT id, workspace, agent_id, created_at, updated_at, messages\n FROM conversations ORDER BY updated_at DESC LIMIT 50`\n );\n return result.rows.map((row) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = row[i];\n // Extract first user message as title\n let firstMsg = \"\";\n try {\n const msgs: ChatMessage[] = JSON.parse(String(obj.messages || \"[]\"));\n const first = msgs.find((m) => m.role === \"user\");\n firstMsg = first ? String(first.content).slice(0, 60) : \"\";\n } catch {}\n return {\n id: obj.id,\n workspace: obj.workspace,\n agent_id: obj.agent_id,\n created_at: obj.created_at,\n updated_at: obj.updated_at,\n title: firstMsg || \"(no messages)\",\n };\n });\n } catch { return []; }\n}\n\nexport async function getConversationById(id: string): Promise<any> {\n try {\n const result = await sqlite.execute(\n \"SELECT id, workspace, agent_id, created_at, updated_at, messages FROM conversations WHERE id = ?\",\n [id]\n );\n if (!result.rows?.length) return null;\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = result.rows[0][i];\n let messages: ChatMessage[] = [];\n try { messages = JSON.parse(String(obj.messages || \"[]\")); } catch {}\n return { id: obj.id, workspace: obj.workspace, agent_id: obj.agent_id,\n created_at: obj.created_at, updated_at: obj.updated_at, messages };\n } catch { return null; }\n}\n\nexport async function deleteConversation(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM conversations WHERE id = ?\", [id]);\n return true;\n } catch { return false; }\n}\n",[],[365,869],"bb0eff04-5b4d-11f1-a9c0-ee650bb23af1","calendar_unified.ts","98d397f8-589e-11f1-a990-ee650bb23af1",511,["D",1780052483377],"// Unified Calendar — merges six event sources (tasks, experiments, objectives,\n// key results, Google Calendar live, custom entries) into a single sorted feed.\n// Every source is wrapped in try/catch so a missing table or down API never\n// blocks the rest of the calendar from loading.\nimport { sqlite } from \"../db.ts\";\nimport { getCredentials } from \"./integrations/_shared.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type CalendarSource =\n | \"tasks\" | \"experiments\" | \"objectives\"\n | \"key_results\" | \"google_calendar\" | \"outlook_calendar\" | \"custom\" | \"nsm\" | \"actions\";\n\nexport interface CalendarEvent {\n id: string; // source-prefixed: \"task:abc\", \"exp_start:xyz\", ...\n source: CalendarSource;\n title: string;\n description?: string;\n date: string; // YYYY-MM-DD (start)\n time?: string; // HH:MM (start) — omitted if all_day\n end_date?: string; // YYYY-MM-DD (end)\n end_time?: string; // HH:MM (end)\n all_day: boolean;\n color: string;\n workspace?: string;\n event_type?: string;\n linked_url?: string; // deep-link back into the originating module\n meta?: Record<string, any>;\n}\n\n// ─── Color / helper maps ─────────────────────────────────────────────────────\n\nconst SOURCE_COLORS: Record<CalendarSource, string> = {\n tasks: \"#3B82F6\", // blue\n experiments: \"#A855F7\", // purple\n objectives: \"#10B981\", // emerald\n key_results: \"#F59E0B\", // amber\n google_calendar: \"#EF4444\", // red\n outlook_calendar:\"#0078D4\", // Outlook blue\n custom: \"#6B7280\", // slate\n nsm: \"#EC4899\", // pink — North Star Metric deadline\n actions: \"#14B8A6\", // teal — AI/agent action items (action_queue)\n};\n\nexport function colorFor(source: CalendarSource): string {\n return SOURCE_COLORS[source] ?? \"#6B7280\";\n}\n\nexport function quarterToEndDate(quarter: string): string | null {\n // Accepts \"2026-Q1\", \"Q1-2026\", \"Q1 2026\", \"Q1/2026\".\n const m = String(quarter).match(/Q?\\s*([1-4]).{0,3}?(\\d{4})|(\\d{4}).{0,3}?Q?\\s*([1-4])/i);\n if (!m) return null;\n const q = parseInt(m[1] || m[4], 10);\n const y = parseInt(m[2] || m[3], 10);\n if (!q || !y) return null;\n switch (q) {\n case 1: return `${y}-03-31`;\n case 2: return `${y}-06-30`;\n case 3: return `${y}-09-30`;\n case 4: return `${y}-12-31`;\n default: return null;\n }\n}\n\nfunction todayISO(): string {\n const d = new Date(); d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\nfunction mondayISO(): string {\n const d = new Date();\n const dow = (d.getDay() + 6) % 7;\n d.setDate(d.getDate() - dow);\n d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\nfunction addDaysISO(iso: string, days: number): string {\n const d = new Date(iso + \"T00:00:00Z\");\n d.setUTCDate(d.getUTCDate() + days);\n return d.toISOString().slice(0, 10);\n}\n\nfunction isoDateOnly(input: any): string | null {\n if (!input) return null;\n const s = String(input);\n // Already YYYY-MM-DD\n if (/^\\d{4}-\\d{2}-\\d{2}$/.test(s)) return s;\n const d = new Date(s);\n if (isNaN(d.getTime())) return null;\n return d.toISOString().slice(0, 10);\n}\n\nfunction isoTime(input: any): string | undefined {\n if (!input) return undefined;\n const s = String(input);\n // HH:MM or HH:MM:SS\n if (/^\\d{2}:\\d{2}(:\\d{2})?$/.test(s)) return s.slice(0, 5);\n const d = new Date(s);\n if (isNaN(d.getTime())) return undefined;\n return d.toISOString().slice(11, 16);\n}\n\nfunction inRange(date: string | null, from: string, to: string): boolean {\n if (!date) return false;\n return date >= from && date <= to;\n}\n\n// ─── Source fetchers ─────────────────────────────────────────────────────────\n\nasync function fetchTasks(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n // tasks table may or may not have a due_date column — query the schema first.\n const cols = await sqlite.execute(\"PRAGMA table_info(tasks)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n const hasDue = colNames.has(\"due_date\");\n if (!hasDue) return [];\n\n const where = [\"due_date IS NOT NULL\", \"date(due_date) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, workspace, status, due_date FROM tasks WHERE ${where.join(\" AND \")} ORDER BY due_date ASC`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const due = isoDateOnly(row[4]);\n if (!due) continue;\n events.push({\n id: `task:${row[0]}`,\n source: \"tasks\",\n title: String(row[1] ?? \"Task\"),\n date: due,\n all_day: true,\n color: colorFor(\"tasks\"),\n workspace: row[2] ?? undefined,\n event_type: \"task_due\",\n linked_url: `/tasks/${row[0]}`,\n meta: { status: row[3] },\n });\n }\n } catch { /* missing table or column → skip */ }\n return events;\n}\n\n// Action items = the action_queue (what the AI/agents prepared, approved, executed). ALL statuses\n// are shown (prepared/approved/executing/done/failed/rejected) — not just completed ones.\nasync function fetchActionItems(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(action_queue)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const where = [\"date(created_at) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace && colNames.has(\"workspace\")) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, integration, endpoint, status, workspace, created_at\n FROM action_queue WHERE ${where.join(\" AND \")} ORDER BY created_at ASC`,\n args,\n );\n for (const row of (r.rows || []) as any[]) {\n const date = isoDateOnly(row[6]);\n if (!date) continue;\n const status = String(row[4] ?? \"prepared\");\n events.push({\n id: `action:${row[0]}`,\n source: \"actions\",\n title: String(row[1] || `${row[2] ?? \"\"} ${row[3] ?? \"\"}`.trim() || \"Action item\"),\n date,\n all_day: true,\n color: colorFor(\"actions\"),\n workspace: row[5] ?? undefined,\n event_type: `action_${status}`,\n linked_url: `/approvals`,\n meta: { status, integration: row[2], endpoint: row[3] },\n });\n }\n } catch { /* missing action_queue → skip */ }\n return events;\n}\n\nasync function fetchExperiments(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n // Probe schema for compatibility — experiments table may have name/title, start_date/end_date.\n const cols = await sqlite.execute(\"PRAGMA table_info(experiments)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"name\") ? \"name\" : colNames.has(\"title\") ? \"title\" : null;\n if (!titleCol || !colNames.has(\"start_date\") || !colNames.has(\"end_date\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const args: any[] = [from, to, from, to];\n if (wsFilter) args.push(workspace);\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, start_date, end_date${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}${colNames.has(\"status\") ? \", status\" : \", NULL\"}\n FROM experiments\n WHERE (date(start_date) BETWEEN date(?) AND date(?))\n OR (date(end_date) BETWEEN date(?) AND date(?))\n ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const id = row[0];\n const title = String(row[1] ?? \"Experiment\");\n const start = isoDateOnly(row[2]);\n const end = isoDateOnly(row[3]);\n const ws = row[4] ?? undefined;\n const status = row[5] ?? undefined;\n if (inRange(start, from, to)) {\n events.push({\n id: `exp_start:${id}`,\n source: \"experiments\",\n title: `Start: ${title}`,\n date: start!,\n all_day: true,\n color: colorFor(\"experiments\"),\n workspace: ws,\n event_type: \"experiment_start\",\n linked_url: `/experiments/${id}`,\n meta: { status },\n });\n }\n if (inRange(end, from, to)) {\n events.push({\n id: `exp_end:${id}`,\n source: \"experiments\",\n title: `End: ${title}`,\n date: end!,\n all_day: true,\n color: colorFor(\"experiments\"),\n workspace: ws,\n event_type: \"experiment_end\",\n linked_url: `/experiments/${id}`,\n meta: { status },\n });\n }\n }\n } catch { /* missing experiments table → skip */ }\n return events;\n}\n\nasync function fetchObjectives(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(objectives)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"title\") ? \"title\" : colNames.has(\"name\") ? \"name\" : null;\n if (!titleCol || !colNames.has(\"quarter\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const args: any[] = [];\n if (wsFilter) args.push(workspace);\n const statusFilter = colNames.has(\"status\") ? \"AND status = 'active'\" : \"\";\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, quarter${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}\n FROM objectives WHERE 1=1 ${statusFilter} ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const quarterEnd = quarterToEndDate(String(row[2] ?? \"\"));\n if (!quarterEnd) continue;\n if (!inRange(quarterEnd, from, to)) continue;\n events.push({\n id: `obj:${row[0]}`,\n source: \"objectives\",\n title: `Objective due: ${row[1]}`,\n date: quarterEnd,\n all_day: true,\n color: colorFor(\"objectives\"),\n workspace: row[3] ?? undefined,\n event_type: \"objective_due\",\n linked_url: `/okrs/${row[0]}`,\n meta: { quarter: row[2] },\n });\n }\n } catch { /* missing objectives table */ }\n return events;\n}\n\nasync function fetchKeyResults(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(key_results)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"title\") ? \"title\" : colNames.has(\"name\") ? \"name\" : null;\n if (!titleCol || !colNames.has(\"updated_at\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const progFilter = colNames.has(\"progress\") ? \"AND progress < 100\" : \"\";\n const args: any[] = [from, to];\n if (wsFilter) args.push(workspace);\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, updated_at${colNames.has(\"progress\") ? \", progress\" : \", 0\"}${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}\n FROM key_results\n WHERE date(updated_at) BETWEEN date(?) AND date(?)\n ${progFilter} ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const updated = isoDateOnly(row[2]);\n if (!updated) continue;\n events.push({\n id: `kr:${row[0]}`,\n source: \"key_results\",\n title: `KR update: ${row[1]}`,\n date: updated,\n all_day: true,\n color: colorFor(\"key_results\"),\n workspace: row[4] ?? undefined,\n event_type: \"kr_update\",\n linked_url: `/okrs?kr=${row[0]}`,\n meta: { progress: Number(row[3]) || 0 },\n });\n }\n } catch { /* missing key_results table */ }\n return events;\n}\n\nasync function fetchGoogleCalendar(from: string, to: string): Promise<CalendarEvent[]> {\n try {\n // Calendar inherits from `google` OAuth credentials; try `calendar` first.\n let creds = await getCredentials(\"calendar\");\n if (!creds?.access_token) creds = await getCredentials(\"google\");\n if (!creds?.access_token) return [];\n\n const timeMin = `${from}T00:00:00Z`;\n const timeMax = `${to}T23:59:59Z`;\n const url = `https://www.googleapis.com/calendar/v3/calendars/primary/events?` +\n `timeMin=${encodeURIComponent(timeMin)}` +\n `&timeMax=${encodeURIComponent(timeMax)}` +\n `&singleEvents=true&orderBy=startTime&maxResults=250`;\n const res = await fetch(url, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return [];\n const data = await res.json().catch(() => ({}));\n const items = Array.isArray(data?.items) ? data.items : [];\n const out: CalendarEvent[] = [];\n for (const ev of items) {\n const startObj = ev.start || {};\n const endObj = ev.end || {};\n const isAllDay = !!startObj.date && !startObj.dateTime;\n const startISO = startObj.dateTime || startObj.date;\n const endISO = endObj.dateTime || endObj.date;\n const date = isoDateOnly(startISO);\n if (!date) continue;\n out.push({\n id: `gcal:${ev.id}`,\n source: \"google_calendar\",\n title: String(ev.summary || \"(no title)\"),\n description: ev.description ? String(ev.description) : undefined,\n date,\n time: isAllDay ? undefined : isoTime(startISO),\n end_date: isoDateOnly(endISO) ?? undefined,\n end_time: isAllDay ? undefined : isoTime(endISO),\n all_day: isAllDay,\n color: colorFor(\"google_calendar\"),\n event_type: \"google_event\",\n linked_url: ev.htmlLink ? String(ev.htmlLink) : undefined,\n meta: { hangoutLink: ev.hangoutLink, location: ev.location, attendees: (ev.attendees || []).length },\n });\n }\n return out;\n } catch {\n return []; // Google down or token expired — entire calendar still loads.\n }\n}\n\nasync function fetchOutlookCalendar(from: string, to: string): Promise<CalendarEvent[]> {\n try {\n // Shares the central `outlook` Microsoft Graph token (same as the Outlook integration).\n const creds = await getCredentials(\"outlook\");\n if (!creds?.access_token) return [];\n const startISO = `${from}T00:00:00Z`;\n const endISO = `${to}T23:59:59Z`;\n const url = `https://graph.microsoft.com/v1.0/me/calendarView?` +\n `startDateTime=${encodeURIComponent(startISO)}&endDateTime=${encodeURIComponent(endISO)}` +\n `&$top=250&$orderby=start/dateTime` +\n `&$select=subject,start,end,isAllDay,webLink,location,organizer`;\n const res = await fetch(url, {\n headers: { Authorization: `Bearer ${creds.access_token}`, Prefer: 'outlook.timezone=\"UTC\"' },\n });\n if (!res.ok) return [];\n const data = await res.json().catch(() => ({}));\n const items = Array.isArray(data?.value) ? data.value : [];\n const out: CalendarEvent[] = [];\n for (const ev of items) {\n const startStr = ev.start?.dateTime || ev.start?.date;\n const endStr = ev.end?.dateTime || ev.end?.date;\n const date = isoDateOnly(startStr);\n if (!date) continue;\n const isAllDay = !!ev.isAllDay;\n out.push({\n id: `ocal:${ev.id}`,\n source: \"outlook_calendar\",\n title: String(ev.subject || \"(no title)\"),\n date,\n time: isAllDay ? undefined : isoTime(startStr),\n end_date: isoDateOnly(endStr) ?? undefined,\n end_time: isAllDay ? undefined : isoTime(endStr),\n all_day: isAllDay,\n color: colorFor(\"outlook_calendar\"),\n event_type: \"outlook_event\",\n linked_url: ev.webLink ? String(ev.webLink) : undefined,\n meta: { location: ev.location?.displayName, organizer: ev.organizer?.emailAddress?.name },\n });\n }\n return out;\n } catch {\n return []; // Outlook down or token lacks Calendars.Read — rest of the calendar still loads.\n }\n}\n\nasync function fetchNSM(from: string, to: string): Promise<CalendarEvent[]> {\n // Surface the North Star Metric deadline as a single calendar event so the team\n // can see \"KPI target due\" on the calendar. NSM is in app_config; we read the\n // most recent target_date.\n const events: CalendarEvent[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, metric_name, target_value, target_date, current_value, unit\n FROM north_star WHERE target_date IS NOT NULL ORDER BY set_at DESC LIMIT 1`\n );\n const rows = (r.rows || []) as any[];\n if (rows.length === 0) return [];\n const row = rows[0];\n const date = isoDateOnly(row[3]);\n if (!date || !inRange(date, from, to)) return [];\n const name = String(row[1] ?? \"North Star Metric\");\n const target = row[2];\n const unit = row[5] ? String(row[5]) : \"\";\n events.push({\n id: `nsm:${row[0]}`,\n source: \"nsm\",\n title: `NSM target due: ${name} → ${target}${unit ? \" \" + unit : \"\"}`,\n date,\n all_day: true,\n color: colorFor(\"nsm\"),\n event_type: \"nsm_target\",\n linked_url: `/command`,\n meta: { current: row[4], target },\n });\n } catch { /* nsm_history missing → skip */ }\n return events;\n}\n\nasync function fetchCustom(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const where = [\"date(event_date) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence\n FROM calendar_entries WHERE ${where.join(\" AND \")} ORDER BY event_date ASC, event_time ASC`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const date = isoDateOnly(row[3]);\n if (!date) continue;\n const allDay = Number(row[10]) === 1;\n events.push({\n id: `cal:${row[0]}`,\n source: \"custom\",\n title: String(row[1] ?? \"Event\"),\n description: row[2] ?? undefined,\n date,\n time: allDay ? undefined : isoTime(row[4]),\n end_date: isoDateOnly(row[5]) ?? undefined,\n end_time: allDay ? undefined : isoTime(row[6]),\n all_day: allDay,\n color: row[9] ? String(row[9]) : colorFor(\"custom\"),\n workspace: row[8] ?? undefined,\n event_type: row[7] ? String(row[7]) : \"custom\",\n meta: { recurrence: row[11] ?? null },\n });\n }\n } catch { /* missing calendar_entries → skip */ }\n return events;\n}\n\n// ─── Unified entry point ─────────────────────────────────────────────────────\n\nexport async function getUnifiedCalendar(opts: {\n from: string;\n to: string;\n workspace?: string;\n sources?: CalendarSource[];\n}): Promise<CalendarEvent[]> {\n const from = opts.from;\n const to = opts.to;\n const all: CalendarSource[] = [\"tasks\", \"experiments\", \"objectives\", \"key_results\", \"google_calendar\", \"outlook_calendar\", \"custom\", \"nsm\", \"actions\"];\n const wanted = new Set(opts.sources?.length ? opts.sources : all);\n\n const tasks: Promise<CalendarEvent[]>[] = [];\n if (wanted.has(\"actions\")) tasks.push(fetchActionItems(from, to, opts.workspace));\n if (wanted.has(\"tasks\")) tasks.push(fetchTasks(from, to, opts.workspace));\n if (wanted.has(\"experiments\")) tasks.push(fetchExperiments(from, to, opts.workspace));\n if (wanted.has(\"objectives\")) tasks.push(fetchObjectives(from, to, opts.workspace));\n if (wanted.has(\"key_results\")) tasks.push(fetchKeyResults(from, to, opts.workspace));\n if (wanted.has(\"google_calendar\")) tasks.push(fetchGoogleCalendar(from, to));\n if (wanted.has(\"outlook_calendar\")) tasks.push(fetchOutlookCalendar(from, to));\n if (wanted.has(\"custom\")) tasks.push(fetchCustom(from, to, opts.workspace));\n if (wanted.has(\"nsm\")) tasks.push(fetchNSM(from, to));\n\n const results = await Promise.allSettled(tasks);\n const merged: CalendarEvent[] = [];\n for (const r of results) {\n if (r.status === \"fulfilled\" && Array.isArray(r.value)) merged.push(...r.value);\n }\n\n merged.sort((a, b) => {\n if (a.date !== b.date) return a.date < b.date ? -1 : 1;\n const at = a.time || \"00:00\";\n const bt = b.time || \"00:00\";\n return at < bt ? -1 : at > bt ? 1 : 0;\n });\n return merged;\n}\n\nexport async function getToday(opts?: { workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const t = todayISO();\n return getUnifiedCalendar({ from: t, to: t, workspace: opts?.workspace, sources: opts?.sources });\n}\n\nexport async function getWeek(opts?: { workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const start = mondayISO();\n const end = addDaysISO(start, 6);\n return getUnifiedCalendar({ from: start, to: end, workspace: opts?.workspace, sources: opts?.sources });\n}\n\nexport async function getUpcoming(opts?: { days?: number; workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const days = opts?.days ?? 14;\n const start = todayISO();\n const end = addDaysISO(start, days);\n return getUnifiedCalendar({ from: start, to: end, workspace: opts?.workspace, sources: opts?.sources });\n}\n\n// ─── CRUD on calendar_entries ────────────────────────────────────────────────\n\nexport interface CalendarEntry {\n id: string;\n title: string;\n description: string | null;\n event_date: string;\n event_time: string | null;\n end_date: string | null;\n end_time: string | null;\n event_type: string | null;\n workspace: string | null;\n color: string | null;\n all_day: number;\n recurrence: string | null;\n created_at: string;\n}\n\nfunction rowToEntry(columns: string[], row: any[]): CalendarEntry {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n event_date: String(obj.event_date),\n event_time: obj.event_time ?? null,\n end_date: obj.end_date ?? null,\n end_time: obj.end_time ?? null,\n event_type: obj.event_type ?? null,\n workspace: obj.workspace ?? null,\n color: obj.color ?? null,\n all_day: Number(obj.all_day) || 0,\n recurrence: obj.recurrence ?? null,\n created_at: String(obj.created_at ?? new Date().toISOString()),\n };\n}\n\nexport async function createCalendarEntry(data: Partial<CalendarEntry>): Promise<CalendarEntry> {\n // QA fix: was passing a UUID string into INTEGER PRIMARY KEY AUTOINCREMENT —\n // SQLITE_MISMATCH datatype mismatch crashed every entry create. Omit id and use\n // last_insert_rowid() to read back the assigned numeric id.\n const now = new Date().toISOString();\n const eventDate = isoDateOnly(data.event_date) ?? todayISO();\n await sqlite.execute(\n `INSERT INTO calendar_entries\n (title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled\",\n data.description ?? null,\n eventDate,\n data.event_time ?? null,\n isoDateOnly(data.end_date) ?? null,\n data.end_time ?? null,\n data.event_type ?? null,\n data.workspace ?? null,\n data.color ?? null,\n data.all_day ? 1 : 0,\n data.recurrence ?? null,\n now,\n ]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((idRow.rows[0] as any[])[0]);\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) {\n return {\n id, title: data.title ?? \"Untitled\", description: data.description ?? null,\n event_date: eventDate, event_time: data.event_time ?? null,\n end_date: isoDateOnly(data.end_date) ?? null, end_time: data.end_time ?? null,\n event_type: data.event_type ?? null, workspace: data.workspace ?? null,\n color: data.color ?? null, all_day: data.all_day ? 1 : 0,\n recurrence: data.recurrence ?? null, created_at: now,\n };\n }\n return rowToEntry(r.columns, r.rows[0] as any[]);\n}\n\nexport async function updateCalendarEntry(id: string, data: Partial<CalendarEntry>): Promise<CalendarEntry | null> {\n const allowed: Array<keyof CalendarEntry> = [\n \"title\", \"description\", \"event_date\", \"event_time\", \"end_date\", \"end_time\",\n \"event_type\", \"workspace\", \"color\", \"all_day\", \"recurrence\",\n ];\n const sets: string[] = [];\n const args: any[] = [];\n for (const k of allowed) {\n if ((data as any)[k] === undefined) continue;\n sets.push(`${k} = ?`);\n let v: any = (data as any)[k];\n if (k === \"event_date\" || k === \"end_date\") v = isoDateOnly(v);\n if (k === \"all_day\") v = v ? 1 : 0;\n args.push(v);\n }\n if (sets.length === 0) {\n const cur = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!cur.rows?.length) return null;\n return rowToEntry(cur.columns, cur.rows[0] as any[]);\n }\n args.push(id);\n await sqlite.execute(`UPDATE calendar_entries SET ${sets.join(\", \")} WHERE id = ?`, args);\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) return null;\n return rowToEntry(r.columns, r.rows[0] as any[]);\n}\n\nexport async function deleteCalendarEntry(id: string): Promise<boolean> {\n await sqlite.execute(\"DELETE FROM calendar_entries WHERE id = ?\", [id]);\n return true;\n}\n\n// Clone a task into a calendar_entries row — lets users \"pin\" a task to the calendar\n// independently of due_date changes on the source.\nexport async function cloneTaskAsEntry(task_id: string): Promise<CalendarEntry | null> {\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(tasks)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n const hasDue = colNames.has(\"due_date\");\n const hasDesc = colNames.has(\"description\");\n const selCols = [\"id\", \"title\", \"workspace\"];\n if (hasDue) selCols.push(\"due_date\"); else selCols.push(\"NULL AS due_date\");\n if (hasDesc) selCols.push(\"description\"); else selCols.push(\"NULL AS description\");\n const r = await sqlite.execute(\n `SELECT ${selCols.join(\", \")} FROM tasks WHERE id = ?`,\n [task_id]\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any[];\n const due = isoDateOnly(row[3]) ?? todayISO();\n return await createCalendarEntry({\n title: String(row[1] ?? \"Task\"),\n description: row[4] ?? null,\n event_date: due,\n workspace: row[2] ?? null,\n event_type: \"task_clone\",\n all_day: 1,\n });\n } catch { return null; }\n}\n\n// ─── LLM agenda briefing ─────────────────────────────────────────────────────\n\nfunction withTimeout<T>(p: Promise<T>, ms: number): Promise<T | null> {\n return new Promise<T | null>((resolve) => {\n let done = false;\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null); } }, ms);\n p.then(\n (v) => { if (!done) { done = true; clearTimeout(timer); resolve(v); } },\n () => { if (!done) { done = true; clearTimeout(timer); resolve(null); } },\n );\n });\n}\n\nfunction deterministicSummary(events: CalendarEvent[], date: string): string {\n if (events.length === 0) return `${date}: no scheduled events. Use the open block to ship one needle-mover.`;\n const counts = new Map<string, number>();\n for (const e of events) counts.set(e.source, (counts.get(e.source) || 0) + 1);\n const parts = Array.from(counts.entries()).map(([s, n]) => `${n} ${s.replace(\"_\", \" \")}`);\n const headline = events[0];\n return `${date}: ${events.length} item${events.length > 1 ? \"s\" : \"\"} (${parts.join(\", \")}). First up: ${headline.title}${headline.time ? ` at ${headline.time}` : \"\"}.`;\n}\n\nexport async function getCalendarSummary(date: string): Promise<{ date: string; summary: string; event_count: number }> {\n const day = isoDateOnly(date) ?? todayISO();\n const events = await getUnifiedCalendar({ from: day, to: day });\n const fallback = deterministicSummary(events, day);\n\n try {\n const cfg = await getLLMConfig();\n if (!cfg?.api_key) return { date: day, summary: fallback, event_count: events.length };\n\n const compact = events.map((e) => ({\n title: e.title, source: e.source, time: e.time, all_day: e.all_day, type: e.event_type,\n }));\n const prompt = [\n \"You are an executive assistant. Produce a 2-3 sentence agenda briefing for the day.\",\n \"Highlight time-sensitive items first, then context-switching cost, then one focus recommendation.\",\n \"Plain prose, no lists, under 350 chars total.\",\n \"\",\n `DATE: ${day}`,\n `EVENTS: ${JSON.stringify(compact)}`,\n ].join(\"\\n\");\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 300,\n temperature: 0.3,\n systemPrompt: \"You write punchy executive briefings.\",\n });\n const result = await withTimeout(llmCall, 10000) as { content?: string } | null;\n const summary = result?.content?.trim() || fallback;\n return { date: day, summary, event_count: events.length };\n } catch {\n return { date: day, summary: fallback, event_count: events.length };\n }\n}\n",[],[365,878],"23580344-5d6a-11f1-8689-1607ee4eb77e","aio.ts","31d1fe3c-5821-11f1-8cb6-ee650bb23af1",634,["D",1780284586582],"// AIO math layer — Phase 5 / Group D:\n// D1: cohort survival LTV (customer_cohorts table)\n// D3: lead engagement velocity (lead_state_changes table)\n// D5: channel attribution (utm_conversions table)\n// Kept separate from intelligence.ts so the math engine file stays focused on funnel/correlations.\nimport { sqlite } from \"../db.ts\";\n\n// ─── D1: COHORT-SURVIVAL LTV ─────────────────────────────────────────────────\n// retention_curve = [m1, m2, m3, ...] retained counts. LTV per customer is the geometric\n// sum of monthly ARPU weighted by the retention rate at each month:\n// LTV = ARPU × Σ (retained_m / cohort_size)\n// Falls back to ARPU / churn_rate when only a single retention value is provided.\n\nexport interface CustomerCohort {\n id: string;\n cohort_label: string;\n cohort_size: number;\n arpu: number;\n retention_curve: number[];\n created_at: string;\n}\n\nfunction rowToCohort(columns: string[], row: any[]): CustomerCohort {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n let curve: number[] = [];\n try { curve = JSON.parse(String(obj.retention_curve || \"[]\")); } catch { curve = []; }\n return {\n id: obj.id,\n cohort_label: obj.cohort_label,\n cohort_size: Number(obj.cohort_size) || 0,\n arpu: Number(obj.arpu) || 0,\n retention_curve: Array.isArray(curve) ? curve.map((n) => Number(n) || 0) : [],\n created_at: obj.created_at,\n };\n}\n\nexport async function listCohorts(): Promise<CustomerCohort[]> {\n const r = await sqlite.execute(\"SELECT id, cohort_label, cohort_size, arpu, retention_curve, created_at FROM customer_cohorts ORDER BY created_at DESC\");\n return (r.rows || []).map((row) => rowToCohort(r.columns, row as any[]));\n}\n\nexport async function createCohort(data: Partial<CustomerCohort>): Promise<CustomerCohort> {\n const id = crypto.randomUUID();\n // QA fix: accept common alias keys (name/label, size/users, ltv/arpu) so the chat tool +\n // UI forms that send shorthand don't silently end up with \"Untitled\" 0/0/0 rows.\n const cohort_label = (data as any).cohort_label || (data as any).name || (data as any).label || \"Untitled\";\n const cohort_size = Number((data as any).cohort_size ?? (data as any).size ?? (data as any).users ?? 0) || 0;\n let arpu = Number((data as any).arpu ?? 0) || 0;\n // If the caller sent an LTV value but no ARPU, back-derive ARPU = LTV / retention_months\n // (default 12) — better than dropping the input.\n if (!arpu && (data as any).ltv) {\n const months = Number((data as any).retention_months ?? 12) || 12;\n arpu = Number((data as any).ltv) / months;\n }\n const curve = JSON.stringify(Array.isArray(data.retention_curve) ? data.retention_curve : []);\n await sqlite.execute(\n \"INSERT INTO customer_cohorts (id, cohort_label, cohort_size, arpu, retention_curve) VALUES (?, ?, ?, ?, ?)\",\n [id, cohort_label, cohort_size, arpu, curve]\n );\n const fetched = await sqlite.execute(\"SELECT id, cohort_label, cohort_size, arpu, retention_curve, created_at FROM customer_cohorts WHERE id = ?\", [id]);\n // QA fix: guard against empty row set so a write that doesn't round-trip returns the\n // intended insert instead of crashing on rows[0].\n if (!fetched.rows?.length) {\n return {\n id,\n cohort_label,\n cohort_size,\n arpu,\n retention_curve: Array.isArray(data.retention_curve) ? data.retention_curve : [],\n created_at: new Date().toISOString(),\n };\n }\n return rowToCohort(fetched.columns, fetched.rows[0] as any[]);\n}\n\nexport async function deleteCohort(id: string): Promise<boolean> {\n await sqlite.execute(\"DELETE FROM customer_cohorts WHERE id = ?\", [id]);\n return true;\n}\n\nexport function computeCohortLTV(cohort: CustomerCohort): number {\n if (!cohort.cohort_size || cohort.cohort_size <= 0 || !cohort.arpu) return 0;\n if (cohort.retention_curve.length === 0) return 0;\n // Geometric sum: LTV = ARPU × Σ (retained_m / cohort_size)\n const retentionRates = cohort.retention_curve.map((r) => r / cohort.cohort_size);\n const survival = retentionRates.reduce((a, b) => a + b, 0);\n return cohort.arpu * survival;\n}\n\nexport async function getCohortSummary(): Promise<{\n cohorts: Array<CustomerCohort & { ltv: number; survival: number }>;\n blended_ltv: number;\n total_customers: number;\n}> {\n const cohorts = await listCohorts();\n const enriched = cohorts.map((c) => {\n const ltv = computeCohortLTV(c);\n const survival = c.retention_curve.length\n ? c.retention_curve.reduce((a, b) => a + b, 0) / (c.cohort_size * c.retention_curve.length)\n : 0;\n return { ...c, ltv, survival };\n });\n const totalCustomers = enriched.reduce((a, c) => a + c.cohort_size, 0);\n const blendedLtv = totalCustomers > 0\n ? enriched.reduce((a, c) => a + c.ltv * c.cohort_size, 0) / totalCustomers\n : 0;\n return { cohorts: enriched, blended_ltv: blendedLtv, total_customers: totalCustomers };\n}\n\n// ─── D3: LEAD ENGAGEMENT VELOCITY ────────────────────────────────────────────\n// Each lead_state_changes row captures a single state transition. Velocity =\n// avg days between MQL and SQL transitions, computed lead-by-lead then averaged.\n\nexport async function recordLeadStateChange(\n lead_email: string,\n to_state: \"cold\" | \"MQL\" | \"SQL\",\n from_state?: \"cold\" | \"MQL\" | \"SQL\",\n): Promise<void> {\n // Skip the insert when the lead is already in this state — avoids noisy duplicate rows\n // every time getLeadScores re-runs and re-classifies the same lead.\n const last = await sqlite.execute(\n \"SELECT to_state FROM lead_state_changes WHERE lead_email = ? ORDER BY changed_at DESC LIMIT 1\",\n [lead_email]\n );\n if (last.rows?.length) {\n const lastState = String((last.rows[0] as any[])[0]);\n if (lastState === to_state) return;\n }\n await sqlite.execute(\n \"INSERT INTO lead_state_changes (lead_email, from_state, to_state) VALUES (?, ?, ?)\",\n [lead_email, from_state ?? null, to_state]\n );\n}\n\nexport async function getEngagementVelocity(): Promise<{\n avg_days_mql_to_sql: number | null;\n avg_days_cold_to_mql: number | null;\n sample_size: number;\n}> {\n const r = await sqlite.execute(\n `SELECT lead_email, to_state, changed_at FROM lead_state_changes ORDER BY lead_email, changed_at ASC`\n );\n if (!r.rows?.length) return { avg_days_mql_to_sql: null, avg_days_cold_to_mql: null, sample_size: 0 };\n\n type Tx = { email: string; state: string; at: number };\n const txs: Tx[] = (r.rows as any[]).map((row) => ({\n email: String(row[0]),\n state: String(row[1]),\n at: new Date(String(row[2])).getTime(),\n }));\n\n const byEmail = new Map<string, Tx[]>();\n for (const t of txs) {\n if (!byEmail.has(t.email)) byEmail.set(t.email, []);\n byEmail.get(t.email)!.push(t);\n }\n\n const mqlToSqlDays: number[] = [];\n const coldToMqlDays: number[] = [];\n for (const [, hist] of byEmail) {\n const mqlTx = hist.find((t) => t.state === \"MQL\");\n const sqlTx = hist.find((t) => t.state === \"SQL\");\n const coldTx = hist.find((t) => t.state === \"cold\");\n if (mqlTx && sqlTx && sqlTx.at > mqlTx.at) {\n mqlToSqlDays.push((sqlTx.at - mqlTx.at) / 86400000);\n }\n if (coldTx && mqlTx && mqlTx.at > coldTx.at) {\n coldToMqlDays.push((mqlTx.at - coldTx.at) / 86400000);\n }\n }\n\n const avg = (arr: number[]) => arr.length ? arr.reduce((a, b) => a + b, 0) / arr.length : null;\n return {\n avg_days_mql_to_sql: avg(mqlToSqlDays),\n avg_days_cold_to_mql: avg(coldToMqlDays),\n sample_size: byEmail.size,\n };\n}\n\n// ─── D5: UTM ATTRIBUTION ─────────────────────────────────────────────────────\n\nexport interface UtmConversion {\n id: number;\n utm_source: string | null;\n utm_medium: string | null;\n utm_campaign: string | null;\n conversion_value: number;\n converted_at: string;\n}\n\nexport async function recordConversion(data: Partial<UtmConversion>): Promise<void> {\n await sqlite.execute(\n \"INSERT INTO utm_conversions (utm_source, utm_medium, utm_campaign, conversion_value) VALUES (?, ?, ?, ?)\",\n [data.utm_source ?? null, data.utm_medium ?? null, data.utm_campaign ?? null, data.conversion_value ?? 0]\n );\n}\n\nexport async function getAttributionSummary(): Promise<{\n channels: Array<{ source: string; conversions: number; value: number; share: number }>;\n total_conversions: number;\n total_value: number;\n window_days: 30;\n}> {\n const r = await sqlite.execute(\n `SELECT COALESCE(utm_source,'direct') AS source, COUNT(*) AS n, COALESCE(SUM(conversion_value), 0) AS v\n FROM utm_conversions\n WHERE converted_at >= datetime('now','-30 days')\n GROUP BY source\n ORDER BY n DESC`\n );\n const rows = (r.rows || []) as any[];\n const channels = rows.map((row) => ({\n source: String(row[0]),\n conversions: Number(row[1]) || 0,\n value: Number(row[2]) || 0,\n share: 0,\n }));\n const totalConv = channels.reduce((a, c) => a + c.conversions, 0);\n const totalVal = channels.reduce((a, c) => a + c.value, 0);\n for (const c of channels) c.share = totalConv > 0 ? c.conversions / totalConv : 0;\n return {\n channels,\n total_conversions: totalConv,\n total_value: totalVal,\n window_days: 30,\n };\n}\n",[],[365,886],"1e29d052-5999-11f1-9cb3-ee650bb23af1","agents.ts","156bf142-56ca-11f1-8a9a-ee650bb23af1",395,["D",1779864959591],"import { sqlite } from \"../db.ts\";\nimport type { Agent } from \"../types.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToAgent(columns: string[], row: any[]): Agent {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n return {\n id: obj.id,\n name: obj.name,\n workspace: obj.workspace,\n description: obj.description ?? \"\",\n system_prompt: obj.system_prompt ?? \"\",\n tools: obj.tools ?? \"[]\",\n model: obj.model ?? \"claude-sonnet-4-6\",\n enabled: obj.enabled ?? 1,\n created_at: obj.created_at,\n updated_at: obj.updated_at,\n created_by: obj.created_by ?? \"user\",\n // B2: budget cap + current spend joined from agent_budgets\n monthly_cap: obj.monthly_cap != null ? Number(obj.monthly_cap) : 250,\n current_spend: obj.current_spend != null ? Number(obj.current_spend) : 0,\n // I3: agent role (drives specialized system prompt prefix server-side)\n role: (obj.role as any) ?? \"general\",\n // Campaign fields — written by create_campaign and the Campaigns page.\n channel: obj.channel ?? null,\n goal: obj.goal ?? null,\n audience: obj.audience ?? null,\n status: obj.status ?? \"active\",\n } as Agent;\n}\n\nasync function upsertBudget(agentId: string, monthly_cap: number): Promise<void> {\n // INSERT OR REPLACE would reset current_spend; do upsert-without-clobber instead.\n const existing = await sqlite.execute(\"SELECT agent_id FROM agent_budgets WHERE agent_id = ?\", [agentId]);\n if (existing.rows?.length) {\n await sqlite.execute(\n \"UPDATE agent_budgets SET monthly_cap = ?, updated_at = datetime('now') WHERE agent_id = ?\",\n [monthly_cap, agentId]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO agent_budgets (agent_id, monthly_cap, current_spend, period_start, updated_at) VALUES (?, ?, 0, datetime('now'), datetime('now'))\",\n [agentId, monthly_cap]\n );\n }\n}\n\n// ─── CRUD functions ───────────────────────────────────────────────────────────\n\nexport async function listAgents(workspace?: string): Promise<Agent[]> {\n let query: string;\n let args: any[];\n\n // B2: LEFT JOIN agent_budgets so monthly_cap/current_spend ride along\n // I3: role column carries the agent's specialized role (general/listener/topic/creator/analyst)\n const base = `SELECT a.id, a.name, a.workspace, a.description, a.system_prompt, a.tools, a.model,\n a.enabled, a.created_at, a.updated_at, a.created_by, a.role,\n a.channel, a.goal, a.audience, a.status,\n b.monthly_cap, b.current_spend\n FROM agents a\n LEFT JOIN agent_budgets b ON b.agent_id = a.id`;\n if (workspace) {\n query = base + \" WHERE a.workspace = ? ORDER BY a.name ASC\";\n args = [workspace];\n } else {\n query = base + \" ORDER BY a.name ASC\";\n args = [];\n }\n\n const result = await sqlite.execute(query, args);\n\n if (!result.rows || result.rows.length === 0) {\n return [];\n }\n\n return result.rows.map((row) => rowToAgent(result.columns, row));\n}\n\nexport async function getAgent(id: string): Promise<Agent | null> {\n const result = await sqlite.execute(\n `SELECT a.id, a.name, a.workspace, a.description, a.system_prompt, a.tools, a.model,\n a.enabled, a.created_at, a.updated_at, a.created_by, a.role,\n a.channel, a.goal, a.audience, a.status,\n b.monthly_cap, b.current_spend\n FROM agents a\n LEFT JOIN agent_budgets b ON b.agent_id = a.id\n WHERE a.id = ?`,\n [id]\n );\n\n if (!result.rows || result.rows.length === 0) {\n return null;\n }\n\n return rowToAgent(result.columns, result.rows[0]);\n}\n\nexport async function createAgent(data: Partial<Agent>): Promise<Agent> {\n const id = crypto.randomUUID();\n const now = new Date().toISOString();\n\n const name = data.name ?? \"Unnamed Agent\";\n const workspace = data.workspace ?? \"marketing\";\n const description = data.description ?? null;\n const system_prompt = data.system_prompt ?? null;\n const tools = (() => {\n if (!data.tools) return \"[]\";\n if (typeof data.tools === \"string\") return data.tools;\n return JSON.stringify(data.tools);\n })();\n const model = (data as any).model ?? \"claude-sonnet-4-6\";\n const created_by = (data as any).created_by ?? \"user\";\n // QA fix: agent role was being silently dropped on create — the role column existed in\n // the schema and chat.ts read it for AGENT_ROLE_PROMPTS injection, but createAgent never\n // wrote it. Listener/Topic/Creator/Analyst roles were always defaulting to 'general'.\n const role = ((data as any).role && String((data as any).role).trim()) || \"general\";\n const channel = (data as any).channel ?? null;\n const goal = (data as any).goal ?? null;\n const audience = (data as any).audience ?? null;\n const status = (data as any).status ?? \"active\";\n\n await sqlite.execute(\n `INSERT INTO agents (id, name, workspace, description, system_prompt, tools, model, enabled, created_at, updated_at, created_by, role, channel, goal, audience, status)\n VALUES (?, ?, ?, ?, ?, ?, ?, 1, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [id, name, workspace, description, system_prompt, tools, model, now, now, created_by, role, channel, goal, audience, status]\n );\n\n // B2: seed budget row (default $250 cap unless caller supplied a monthly_cap)\n const monthly_cap = (data as any).monthly_cap != null && Number.isFinite(Number((data as any).monthly_cap))\n ? Number((data as any).monthly_cap)\n : 250;\n await upsertBudget(id, monthly_cap);\n\n const created = await getAgent(id);\n if (!created) {\n throw new Error(`Failed to retrieve agent after creation (id=${id})`);\n }\n return created;\n}\n\nexport async function updateAgent(\n id: string,\n data: Partial<Agent>\n): Promise<Agent | null> {\n const existing = await getAgent(id);\n if (!existing) {\n return null;\n }\n\n const now = new Date().toISOString();\n\n const name = data.name ?? existing.name;\n const workspace = data.workspace ?? existing.workspace;\n const description = \"description\" in data ? (data.description ?? null) : (existing.description ?? null);\n const system_prompt = \"system_prompt\" in data ? (data.system_prompt ?? null) : (existing.system_prompt ?? null);\n const tools = \"tools\" in data\n ? (typeof data.tools === \"string\" ? data.tools : JSON.stringify(data.tools ?? []))\n : (existing.tools ?? \"[]\");\n const model = (data as any).model ?? (existing as any).model ?? \"claude-sonnet-4-6\";\n const enabled = \"enabled\" in data ? (data as any).enabled : (existing.enabled ?? 1);\n // QA fix: persist role on update too (was being silently dropped).\n const role = \"role\" in (data as any)\n ? (((data as any).role && String((data as any).role).trim()) || \"general\")\n : ((existing as any).role ?? \"general\");\n const channel = \"channel\" in (data as any) ? ((data as any).channel ?? null) : ((existing as any).channel ?? null);\n const goal = \"goal\" in (data as any) ? ((data as any).goal ?? null) : ((existing as any).goal ?? null);\n const audience = \"audience\" in (data as any) ? ((data as any).audience ?? null) : ((existing as any).audience ?? null);\n const status = \"status\" in (data as any) ? ((data as any).status ?? \"active\") : ((existing as any).status ?? \"active\");\n\n await sqlite.execute(\n `UPDATE agents SET name = ?, workspace = ?, description = ?, system_prompt = ?, tools = ?, model = ?, enabled = ?, updated_at = ?, role = ?, channel = ?, goal = ?, audience = ?, status = ? WHERE id = ?`,\n [name, workspace, description, system_prompt, tools, model, enabled, now, role, channel, goal, audience, status, id]\n );\n\n // B2: persist monthly_cap when caller supplies one (omit-to-keep semantics).\n if (\"monthly_cap\" in (data as any)) {\n const cap = Number((data as any).monthly_cap);\n if (Number.isFinite(cap) && cap >= 0) await upsertBudget(id, cap);\n }\n\n return getAgent(id);\n}\n\nexport async function deleteAgent(id: string): Promise<boolean> {\n const existing = await getAgent(id);\n if (!existing) {\n return false;\n }\n\n await sqlite.execute(\"DELETE FROM agents WHERE id = ?\", [id]);\n return true;\n}\n\n// ─── C3: Agent execution log ────────────────────────────────────────────────\nexport async function listAgentRuns(agentId: string, limit = 50): Promise<any[]> {\n const result = await sqlite.execute(\n `SELECT id, agent_id, status, trigger, started_at, finished_at, duration_ms, output, error\n FROM agent_runs WHERE agent_id = ? ORDER BY started_at DESC LIMIT ?`,\n [agentId, limit]\n );\n return (result.rows || []).map((row: any) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = row[i];\n return obj;\n });\n}\n\nexport async function startAgentRun(agentId: string, trigger: string, input?: string): Promise<string> {\n const id = crypto.randomUUID();\n await sqlite.execute(\n \"INSERT INTO agent_runs (id, agent_id, status, trigger, started_at, input) VALUES (?, ?, 'running', ?, datetime('now'), ?)\",\n [id, agentId, trigger || 'manual', input ?? null]\n );\n return id;\n}\n\nexport async function finishAgentRun(runId: string, status: 'completed' | 'failed' | 'cancelled', output?: string, error?: string): Promise<void> {\n // Compute duration from started_at recorded server-side to avoid clock drift.\n await sqlite.execute(\n `UPDATE agent_runs\n SET status = ?, finished_at = datetime('now'),\n duration_ms = CAST((julianday('now') - julianday(started_at)) * 86400000 AS INTEGER),\n output = ?, error = ?\n WHERE id = ?`,\n [status, output ?? null, error ?? null, runId]\n );\n}\n",[],[365,894],"51ab9480-5a69-11f1-ad64-ee650bb23af1","actions.ts","51ab957a-5a69-11f1-ad64-ee650bb23af1",429,["D",1779954381335],"// Human-in-the-loop action queue. The AI / agents PREPARE external actions here\n// (send email, publish post, create LinkedIn post, create calendar event). Nothing is\n// sent until a human APPROVES; on approval the real integration call runs (reusing the\n// same handleIntegrationProxy dispatch as the rest of the app), the result is recorded,\n// and any linked task is closed (which propagates to the measure layer).\nimport { sqlite } from \"../db.ts\";\nimport { handleIntegrationProxy, getCredentials } from \"./integrations.ts\";\nimport { updateTask } from \"./tasks.ts\";\n\nexport interface QueuedAction {\n id: number;\n source_type: string;\n source_id: string | null;\n integration: string;\n endpoint: string;\n payload: any;\n preview: string | null;\n title: string | null;\n workspace: string | null;\n status: string; // prepared | approved | executing | done | failed | rejected\n result: any;\n task_id: string | null;\n created_at: string;\n approved_at: string | null;\n executed_at: string | null;\n}\n\nconst COLS = `id, source_type, source_id, integration, endpoint, payload, preview, title,\n workspace, status, result, task_id, created_at, approved_at, executed_at`;\n\nfunction parseMaybe(v: any): any {\n if (v == null) return null;\n try { return JSON.parse(String(v)); } catch { return v; }\n}\n\nfunction rowToAction(cols: string[], row: any[]): QueuedAction {\n const o: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) o[cols[i]] = row[i];\n return {\n id: Number(o.id),\n source_type: String(o.source_type ?? \"chat\"),\n source_id: o.source_id ?? null,\n integration: String(o.integration),\n endpoint: String(o.endpoint),\n payload: parseMaybe(o.payload),\n preview: o.preview ?? null,\n title: o.title ?? null,\n workspace: o.workspace ?? null,\n status: String(o.status ?? \"prepared\"),\n result: parseMaybe(o.result),\n task_id: o.task_id ?? null,\n created_at: String(o.created_at ?? \"\"),\n approved_at: o.approved_at ?? null,\n executed_at: o.executed_at ?? null,\n };\n}\n\nexport async function getAction(id: string | number): Promise<QueuedAction | null> {\n try {\n const r = await sqlite.execute(`SELECT ${COLS} FROM action_queue WHERE id = ?`, [Number(id)]);\n if (!r.rows?.length) return null;\n return rowToAction(r.columns, r.rows[0] as any[]);\n } catch { return null; }\n}\n\nexport async function listActions(status?: string): Promise<QueuedAction[]> {\n try {\n const r = status\n ? await sqlite.execute(`SELECT ${COLS} FROM action_queue WHERE status = ? ORDER BY created_at DESC LIMIT 100`, [status])\n : await sqlite.execute(`SELECT ${COLS} FROM action_queue ORDER BY created_at DESC LIMIT 100`);\n return (r.rows || []).map((row: any) => rowToAction(r.columns, row));\n } catch { return []; }\n}\n\nexport async function enqueueAction(data: {\n integration: string; endpoint: string; payload?: any;\n preview?: string; title?: string; workspace?: string;\n source_type?: string; source_id?: string; task_id?: string;\n}): Promise<QueuedAction | null> {\n if (!data.integration || !data.endpoint) return null;\n await sqlite.execute(\n `INSERT INTO action_queue\n (source_type, source_id, integration, endpoint, payload, preview, title, workspace, task_id, status)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 'prepared')`,\n [\n data.source_type ?? \"chat\", data.source_id ?? null,\n data.integration, data.endpoint,\n JSON.stringify(data.payload ?? {}),\n data.preview ?? null, data.title ?? null,\n data.workspace ?? \"marketing\", data.task_id ?? null,\n ],\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n return await getAction(Number((idRow.rows[0] as any[])[0]));\n}\n\nexport async function rejectAction(id: string | number): Promise<QueuedAction | null> {\n await sqlite.execute(\n \"UPDATE action_queue SET status='rejected' WHERE id = ? AND status IN ('prepared','approved')\",\n [Number(id)],\n );\n return await getAction(id);\n}\n\n// Approve → run the real integration call → record result → close linked task.\nexport async function approveAction(id: string | number): Promise<QueuedAction | null> {\n const action = await getAction(id);\n if (!action) return null;\n if (action.status !== \"prepared\" && action.status !== \"approved\") return action;\n\n await sqlite.execute(\n \"UPDATE action_queue SET status='executing', approved_at=datetime('now') WHERE id = ?\",\n [Number(id)],\n );\n\n let ok = false;\n let result: any = null;\n try {\n const creds = await getCredentials(action.integration);\n if (!creds) {\n result = { error: `Integration '${action.integration}' not configured` };\n } else {\n const params = new URLSearchParams({ endpoint: action.endpoint });\n const req = new Request(\n `https://internal/api/integrations/${action.integration}/data?endpoint=${encodeURIComponent(action.endpoint)}`,\n { method: \"POST\", headers: { \"Content-Type\": \"application/json\" }, body: JSON.stringify(action.payload ?? {}) },\n );\n const res = await handleIntegrationProxy(req, action.integration, params, creds);\n result = await res.json().catch(() => ({}));\n ok = res.ok && !(result && result.error);\n }\n } catch (e: any) {\n result = { error: String(e?.message || e) };\n }\n\n await sqlite.execute(\n \"UPDATE action_queue SET status = ?, result = ?, executed_at = datetime('now') WHERE id = ?\",\n [ok ? \"done\" : \"failed\", JSON.stringify(result ?? {}), Number(id)],\n );\n\n // On success, close the linked task — updateTask triggers propagateTaskDone (measure layer).\n if (ok && action.task_id) {\n try { await updateTask(action.task_id, { status: \"done\" } as any); } catch { /* non-critical */ }\n }\n return await getAction(id);\n}\n",[],[365,902],"c7a5c918-5806-11f1-97e7-ee650bb23af1","c7a5c9fe-5806-11f1-97e7-ee650bb23af1","de234ff5-1325-4b93-9eef-d2295e01092a",120,["D",1779692156770],"// YouTube · channels, videos, playlists, search.\n// Docs: https://developers.google.com/youtube/v3\n// Note: mine=true requires OAuth. API key works only for public endpoints.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/youtube/v3\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const hasOAuth = !!creds.access_token;\n const headers: Record<string, string> = {};\n let authParam = \"\";\n\n if (hasOAuth) headers[\"Authorization\"] = `Bearer ${creds.access_token}`;\n else if (creds.api_key) authParam = `key=${encodeURIComponent(creds.api_key)}`;\n\n const jp = (a: string, b: string) => [a, b].filter(Boolean).join(\"&\");\n\n let url: string;\n switch (endpoint) {\n case \"channels\":\n url = hasOAuth\n ? `${BASE}/channels?part=snippet,statistics,brandingSettings,contentDetails&mine=true&${jp(params, authParam)}`\n : `${BASE}/channels?part=snippet,statistics&chart=mostPopular&maxResults=5&${jp(params, authParam)}`;\n break;\n case \"videos\":\n url = `${BASE}/videos?part=snippet,statistics,contentDetails&chart=mostPopular&maxResults=10&${jp(params, authParam)}`;\n break;\n case \"playlists\":\n if (!hasOAuth) return json({ error: \"Playlists requires OAuth. Connect Google in Settings.\" }, 401);\n url = `${BASE}/playlists?part=snippet,contentDetails&mine=true&maxResults=20&${jp(params, authParam)}`;\n break;\n case \"search\":\n url = `${BASE}/search?part=snippet&type=video&maxResults=10&${jp(params, authParam)}`;\n break;\n default:\n url = `${BASE}/${endpoint}?${jp(params, authParam)}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,568],"integrations","58ebcc96-5b19-11f1-8599-ee650bb23af1","c8f8da6c-5806-11f1-9882-ee650bb23af1",467,["D",1780029984924],"// WordPress REST API · posts, pages, media, categories, publish-post.\n// Docs: https://developer.wordpress.org/rest-api/\n// Auth: Basic base64(username:application_password). Requires HTTPS.\nimport { json, safeJson, isSafePublicUrl, IntegrationCredentials } from \"./_shared.ts\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const siteUrl = (creds.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl) return json({ error: \"WordPress site URL required. Add it in Settings.\" }, 400);\n if (!isSafePublicUrl(siteUrl)) return json({ error: \"WordPress site URL must be a public https:// address.\" }, 400);\n // QA fix: refuse to build malformed Basic auth from empty creds (used to silently 401).\n const username = (creds.api_key ?? \"\").trim();\n const appPassword = (creds.api_secret ?? \"\").trim();\n if (!username || !appPassword) {\n return json({ error: \"WordPress requires both a username (api_key) and an application password (api_secret) in Settings.\" }, 400);\n }\n const base = `${siteUrl}/wp-json/wp/v2`;\n const basicAuth = btoa(`${username}:${appPassword}`);\n const headers = { Authorization: `Basic ${basicAuth}`, \"Content-Type\": \"application/json\" };\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"posts\":\n url = `${base}/posts?per_page=20&orderby=date&order=desc&${params}`;\n break;\n case \"pages\":\n url = `${base}/pages?per_page=20&orderby=date&order=desc&${params}`;\n break;\n case \"media\":\n url = `${base}/media?per_page=20&${params}`;\n break;\n case \"categories\":\n url = `${base}/categories?per_page=50&${params}`;\n break;\n case \"publish-post\":\n url = `${base}/posts`;\n method = \"POST\";\n body = await req.text();\n break;\n default:\n url = `${base}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const siteUrl = (creds.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl) return { ok: false, message: \"Site URL is required\" };\n if (!isSafePublicUrl(siteUrl)) return { ok: false, message: \"Site URL must be a public https:// address\" };\n const username = (creds.api_key ?? \"\").trim();\n const appPassword = (creds.api_secret ?? \"\").trim();\n if (!username || !appPassword) return { ok: false, message: \"Username + application password required\" };\n const basicAuth = btoa(`${username}:${appPassword}`);\n const res = await fetch(`${siteUrl}/wp-json/wp/v2/users/me`, { headers: { Authorization: `Basic ${basicAuth}` } });\n if (!res.ok) return { ok: false, message: `Auth failed (${res.status})` };\n return { ok: true, message: \"Credentials verified\" };\n}\n",[],[365,917,576],"6d858758-5b48-11f1-94b4-ee650bb23af1","6d85882a-5b48-11f1-94b4-ee650bb23af1",499,["D",1780050205835],"// Microsoft Teams (Microsoft Graph) · teams, channels, channel messages, chats, chat messages.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/teams-api-overview\n// READ: teams (joinedTeams), channels, messages, chats, chat-messages\n// WRITE: send-channel, send-chat\n// Shares the central `outlook` Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n // Id-bearing reads + writes accept identifiers from the POST body (the panel form) or the query\n // string. Read the body once and pick from either (query-first).\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"teams\":\n url = `${BASE}/me/joinedTeams`;\n break;\n case \"channels\": {\n const teamId = pick(\"teamId\");\n if (!teamId) return json({ error: \"teamId required\" }, 400);\n url = `${BASE}/teams/${encodeURIComponent(teamId)}/channels`;\n break;\n }\n case \"messages\": {\n const teamId = pick(\"teamId\");\n const channelId = pick(\"channelId\");\n if (!teamId || !channelId) return json({ error: \"teamId and channelId required\" }, 400);\n url = `${BASE}/teams/${encodeURIComponent(teamId)}/channels/${encodeURIComponent(channelId)}/messages?$top=30`;\n break;\n }\n case \"chats\":\n url = `${BASE}/me/chats?$top=30`;\n break;\n case \"chat-messages\": {\n const chatId = pick(\"chatId\");\n if (!chatId) return json({ error: \"chatId required\" }, 400);\n url = `${BASE}/me/chats/${encodeURIComponent(chatId)}/messages?$top=30`;\n break;\n }\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"send-channel\": {\n const teamId = pick(\"teamId\");\n const channelId = pick(\"channelId\");\n if (!teamId || !channelId) return json({ error: \"teamId and channelId required\" }, 400);\n if (!bodyParams.content) return json({ error: \"content required\" }, 400);\n const res = await fetch(`${BASE}/teams/${encodeURIComponent(teamId)}/channels/${encodeURIComponent(channelId)}/messages`, {\n method: \"POST\", headers,\n body: JSON.stringify({ body: { contentType: \"html\", content: String(bodyParams.content) } }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"send-chat\": {\n const chatId = pick(\"chatId\");\n if (!chatId) return json({ error: \"chatId required\" }, 400);\n if (!bodyParams.content) return json({ error: \"content required\" }, 400);\n const res = await fetch(`${BASE}/me/chats/${encodeURIComponent(chatId)}/messages`, {\n method: \"POST\", headers,\n body: JSON.stringify({ body: { content: String(bodyParams.content) } }),\n });\n return json(await safeJson(res), res.status);\n }\n\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(`${BASE}/me/joinedTeams`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,584],"c64033d8-5806-11f1-8516-ee650bb23af1","c64034aa-5806-11f1-8516-ee650bb23af1",117,["D",1779692154414],"// Google Sheets · list spreadsheets, read values, create/append/update.\n// Docs: https://developers.google.com/sheets/api/reference/rest\n// READ: list (via Drive query), values, metadata, batchGet\n// WRITE: create, appendRow, updateCells\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst SHEETS_BASE = \"https://sheets.googleapis.com/v4/spreadsheets\";\nconst DRIVE_BASE = \"https://www.googleapis.com/drive/v3/files\";\nconst FULL_FILES = \"files(id,name,mimeType,modifiedTime,createdTime,size,webViewLink,iconLink,owners(displayName,emailAddress),lastModifyingUser(displayName,emailAddress),shared,starred),nextPageToken\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"list\": {\n const driveSp = new URLSearchParams({\n q: \"mimeType='application/vnd.google-apps.spreadsheet' and trashed=false\",\n pageSize: sp.get(\"pageSize\") || \"50\",\n orderBy: \"modifiedTime desc\",\n fields: FULL_FILES,\n });\n url = `${DRIVE_BASE}?${driveSp.toString()}`;\n break;\n }\n case \"values\": {\n const id = sp.get(\"spreadsheetId\");\n const range = sp.get(\"range\") || \"A1:Z100\";\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n const opt = sp.get(\"valueRenderOption\") || \"FORMATTED_VALUE\";\n url = `${SHEETS_BASE}/${id}/values/${encodeURIComponent(range)}?valueRenderOption=${opt}&dateTimeRenderOption=FORMATTED_STRING`;\n break;\n }\n case \"metadata\": {\n const id = sp.get(\"spreadsheetId\");\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n url = `${SHEETS_BASE}/${id}?includeGridData=false`;\n break;\n }\n case \"batchGet\": {\n const id = sp.get(\"spreadsheetId\");\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n const ranges = sp.getAll(\"ranges\");\n const rangesQ = ranges.length ? ranges.map(r => `ranges=${encodeURIComponent(r)}`).join(\"&\") : \"ranges=A1:Z1000\";\n url = `${SHEETS_BASE}/${id}/values:batchGet?${rangesQ}&valueRenderOption=FORMATTED_VALUE`;\n break;\n }\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"create\": {\n const body = await req.json().catch(() => ({}));\n if (!body.title) return json({ error: \"title required\" }, 400);\n const res = await fetch(SHEETS_BASE, {\n method: \"POST\", headers,\n body: JSON.stringify({ properties: { title: body.title } }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"appendRow\": {\n const body = await req.json().catch(() => ({}));\n if (!body.spreadsheetId) return json({ error: \"spreadsheetId required\" }, 400);\n const range = body.range || \"Sheet1!A:Z\";\n const values = Array.isArray(body.values?.[0]) ? body.values : [body.values || []];\n const res = await fetch(\n `${SHEETS_BASE}/${body.spreadsheetId}/values/${encodeURIComponent(range)}:append?valueInputOption=USER_ENTERED&insertDataOption=INSERT_ROWS`,\n { method: \"POST\", headers, body: JSON.stringify({ values }) }\n );\n return json(await safeJson(res), res.status);\n }\n case \"updateCells\": {\n const body = await req.json().catch(() => ({}));\n if (!body.spreadsheetId || !body.range) return json({ error: \"spreadsheetId + range required\" }, 400);\n const res = await fetch(\n `${SHEETS_BASE}/${body.spreadsheetId}/values/${encodeURIComponent(body.range)}?valueInputOption=USER_ENTERED`,\n { method: \"PUT\", headers, body: JSON.stringify({ values: body.values }) }\n );\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${SHEETS_BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,592],"c96adf54-5806-11f1-8bf1-ee650bb23af1","c96ae012-5806-11f1-8bf1-ee650bb23af1",124,["D",1779692159729],"// Product Hunt · GraphQL API for today's posts, topics.\n// Docs: https://api.producthunt.com/v2/docs\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst URL = \"https://api.producthunt.com/v2/api/graphql\";\n\nexport async function handle(_req: Request, endpoint: string, searchParams: URLSearchParams, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.api_key}`, \"Content-Type\": \"application/json\" };\n\n let query: string;\n switch (endpoint) {\n case \"posts\":\n query = `{ posts(first: 20, order: RANKING) { edges { node { id name tagline votesCount commentsCount url createdAt thumbnail { url } topics { edges { node { name } } } } } } }`;\n break;\n case \"topics\":\n query = `{ topics(first: 30) { edges { node { id name slug followersCount } } } }`;\n break;\n default:\n query = searchParams.get(\"query\") ?? `{ posts(first: 10, order: RANKING) { edges { node { id name tagline votesCount } } } }`;\n }\n\n const res = await fetch(URL, { method: \"POST\", headers, body: JSON.stringify({ query }) });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(URL, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.api_key}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({ query: \"{ viewer { id } }\" }),\n });\n if (!res.ok) return { ok: false, message: \"Invalid API key\" };\n const d = await safeJson(res);\n if (d.errors?.some((e: any) => /auth|unauthorized/i.test(e.message ?? \"\"))) return { ok: false, message: \"Invalid API key\" };\n return { ok: true, message: \"API key verified\" };\n}\n",[],[365,917,600],"255707c2-5d69-11f1-8f15-1607ee4eb77e","c88a5420-5806-11f1-b58f-ee650bb23af1",617,["D",1780284160447],"// Outlook (Microsoft Graph) · mail, calendar, contacts.\n// Docs: https://learn.microsoft.com/en-us/graph/api/overview\n// READ: inbox/messages/calendar/contacts\n// WRITE: send\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"inbox\":\n case \"messages\":\n url = `${BASE}/messages?$top=20&$orderby=receivedDateTime desc&${params}`;\n break;\n case \"send\":\n url = `${BASE}/sendMail`;\n method = \"POST\";\n body = await req.text();\n break;\n case \"calendar\":\n url = `${BASE}/calendar/events?$top=20&$orderby=start/dateTime asc&${params}`;\n break;\n case \"contacts\":\n url = `${BASE}/contacts?$top=20&${params}`;\n break;\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,608],"2502124e-5d69-11f1-ba26-1607ee4eb77e","6c4c2216-5b48-11f1-bc6f-ee650bb23af1",616,["D",1780284159879],"// OneDrive (Microsoft Graph) · browse, search, and create files & folders.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/onedrive\n// READ: recent / files / search WRITE: createFolder\n// Shares the central `outlook` Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me/drive\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n // Form-based actions (search, createFolder) POST a JSON body from the panel; direct/AI calls\n // may pass query params. Read the body once and accept either source (query-first).\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n case \"recent\":\n url = `${BASE}/recent`;\n break;\n case \"files\":\n url = `${BASE}/root/children?$top=100&$orderby=lastModifiedDateTime desc`;\n break;\n case \"search\": {\n const q = String(pick(\"q\") || pick(\"query\") || \"\").replace(/'/g, \"''\");\n url = `${BASE}/root/search(q='${encodeURIComponent(q)}')`;\n break;\n }\n case \"createFolder\": {\n const name = pick(\"name\");\n if (!name) return json({ error: \"name required\" }, 400);\n const res = await fetch(`${BASE}/root/children`, {\n method: \"POST\", headers,\n body: JSON.stringify({ name, folder: {}, \"@microsoft.graph.conflictBehavior\": \"rename\" }),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,616],"24a2afc0-5d69-11f1-8218-1607ee4eb77e","c8228bd8-5806-11f1-a99e-ee650bb23af1",615,["D",1780284159258],16,"// LinkedIn v2 API · profile, posts, connections.\n// Docs: https://learn.microsoft.com/en-us/linkedin/marketing/integrations/community-management/shares/ugc-post-api\n// Most endpoints require Marketing Developer Platform approval.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://api.linkedin.com/v2\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n // QA fix: guard missing access_token instead of sending 'Bearer undefined'\n if (!creds.access_token) {\n return json({ error: \"LinkedIn access token required. Connect LinkedIn under Settings → LinkedIn.\" }, 401);\n }\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const authorId = (creds.extra_config?.author_id ?? \"\").toString().trim();\n\n let url: string;\n switch (endpoint) {\n case \"profile\":\n url = `${BASE}/me?projection=(id,firstName,lastName,profilePicture)`;\n break;\n case \"posts\":\n // QA fix: posts requires an author_id — without it LinkedIn returns a confusing 400.\n if (!authorId) return json({ error: \"LinkedIn 'posts' requires author_id (your urn:li:person:... or urn:li:organization:...). Add it under Settings → LinkedIn.\" }, 400);\n url = `${BASE}/ugcPosts?q=authors&authors=List(${authorId})&${params}`;\n break;\n case \"analytics\":\n url = `${BASE}/organizationalEntityShareStatistics?${params}`;\n break;\n case \"connections\":\n url = `${BASE}/connections?q=viewer&${params}`;\n break;\n // ─── WRITE ────────────────────────────────────────────────────────────────\n case \"create-post\": {\n if (!authorId) return json({ error: \"LinkedIn 'create-post' requires author_id (urn:li:person:... or urn:li:organization:...). Add it under Settings → LinkedIn.\" }, 400);\n const body = await _req.json().catch(() => ({}));\n const text = String(body.text ?? body.content ?? \"\").trim();\n if (!text) return json({ error: \"text required\" }, 400);\n const payload = {\n author: authorId,\n lifecycleState: \"PUBLISHED\",\n specificContent: {\n \"com.linkedin.ugc.ShareContent\": {\n shareCommentary: { text },\n shareMediaCategory: \"NONE\",\n },\n },\n visibility: { \"com.linkedin.ugc.MemberNetworkVisibility\": \"PUBLIC\" },\n };\n const res = await fetch(`${BASE}/ugcPosts`, {\n method: \"POST\",\n headers: { ...headers, \"X-Restli-Protocol-Version\": \"2.0.0\" },\n body: JSON.stringify(payload),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n if (!creds.access_token) return { ok: false, message: \"Access token required\" };\n const res = await fetch(`${BASE}/me`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n\nexport async function exchangeLinkedInCode(code: string, redirectUri: string, clientId: string, clientSecret: string): Promise<string> {\n const res = await fetch(\"https://www.linkedin.com/oauth/v2/accessToken\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({\n grant_type: \"authorization_code\",\n code,\n redirect_uri: redirectUri,\n client_id: clientId,\n client_secret: clientSecret,\n }),\n });\n const data = await res.json() as any;\n if (!res.ok || !data.access_token) {\n throw new Error(data.error_description || data.error || \"LinkedIn token exchange failed\");\n }\n return data.access_token;\n}\n",[],[365,917,624],"a44abc86-5968-11f1-a954-ee650bb23af1","c4da5e06-5806-11f1-b423-ee650bb23af1",311,["D",1779844139288],"// Google Search Console · query, page, and sitemap data.\n// Docs: https://developers.google.com/webmaster-tools/v1/api_reference_index\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://searchconsole.googleapis.com/webmasters/v3/sites\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n // QA fix: guard required fields up front instead of constructing a malformed URL.\n const rawSiteUrl = (creds.extra_config?.site_url ?? \"\").toString().trim();\n if (!rawSiteUrl) {\n return json({ error: \"Search Console site URL required. Add 'Site URL' under Settings → GSC (e.g. https://example.com or sc-domain:example.com).\" }, 400);\n }\n if (!creds.access_token) {\n return json({ error: \"Search Console needs Google OAuth. Click 'Connect with Google' in Settings.\" }, 401);\n }\n const siteUrl = encodeURIComponent(rawSiteUrl);\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"search-analytics\":\n case \"searchanalytics\": {\n url = `${BASE}/${siteUrl}/searchAnalytics/query`;\n method = \"POST\";\n const saDays = parseInt(sp.get(\"days\") || \"90\", 10);\n const saStartDate = new Date(Date.now() - saDays * 86400000).toISOString().slice(0, 10);\n body = (await req.text()) || JSON.stringify({\n startDate: saStartDate,\n endDate: new Date().toISOString().slice(0, 10),\n dimensions: [\"query\"],\n rowLimit: 50,\n });\n break;\n }\n case \"pages\": {\n url = `${BASE}/${siteUrl}/searchAnalytics/query`;\n method = \"POST\";\n const pgDays = parseInt(sp.get(\"days\") || \"90\", 10);\n const pgStartDate = new Date(Date.now() - pgDays * 86400000).toISOString().slice(0, 10);\n body = JSON.stringify({\n startDate: pgStartDate,\n endDate: new Date().toISOString().slice(0, 10),\n dimensions: [\"page\"],\n rowLimit: 25,\n });\n break;\n }\n case \"sitemaps\":\n url = `${BASE}/${siteUrl}/sitemaps`;\n break;\n default:\n url = `${BASE}/${siteUrl}/${endpoint}`;\n }\n\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,632],"c54abd2c-5806-11f1-8bf1-ee650bb23af1","c54abdfe-5806-11f1-8bf1-ee650bb23af1",115,["D",1779692152815],"// Gmail · inbox + drafts + send + label management.\n// Docs: https://developers.google.com/gmail/api/reference/rest\n// READ: inbox/unread/sent/starred/labels/profile/threads/message?id=X\n// WRITE: send, createDraft, modifyLabels\n// List endpoints auto-hydrate the first 20 messages with sender/subject/date/labelIds/snippet/size.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://gmail.googleapis.com/gmail/v1/users/me\";\nconst STD_HEADERS = [\"From\", \"To\", \"Cc\", \"Bcc\", \"Subject\", \"Date\", \"Reply-To\", \"Message-ID\"];\n\n// RFC 2822 → base64url for Gmail's `raw` param.\nfunction encodeRaw(headers: string[], body: string): string {\n return btoa(unescape(encodeURIComponent([...headers, \"Content-Type: text/plain; charset=UTF-8\", \"\", body].join(\"\\r\\n\"))))\n .replace(/\\+/g, \"-\").replace(/\\//g, \"_\").replace(/=+$/, \"\");\n}\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n const HYDRATE_LIMIT = parseInt(sp.get(\"hydrate\") || \"20\", 10);\n sp.delete(\"hydrate\");\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"inbox\":\n case \"messages\":\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n if (!sp.has(\"q\")) sp.set(\"q\", \"in:inbox\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"threads\":\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/threads?${sp.toString()}`;\n break;\n case \"labels\":\n url = `${BASE}/labels`;\n break;\n case \"profile\":\n url = `${BASE}/profile`;\n break;\n case \"unread\":\n sp.set(\"q\", \"is:unread in:inbox\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"sent\":\n sp.set(\"q\", \"in:sent\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"starred\":\n sp.set(\"q\", \"is:starred\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"send\": {\n const body = await req.json().catch(() => ({}));\n if (!body.to || !body.subject) return json({ error: \"to + subject required\" }, 400);\n const lines = [`To: ${body.to}`];\n if (body.cc) lines.push(`Cc: ${body.cc}`);\n if (body.bcc) lines.push(`Bcc: ${body.bcc}`);\n lines.push(`Subject: ${body.subject}`);\n const raw = encodeRaw(lines, body.body || \"\");\n const res = await fetch(`${BASE}/messages/send`, { method: \"POST\", headers, body: JSON.stringify({ raw }) });\n return json(await safeJson(res), res.status);\n }\n case \"createDraft\": {\n const body = await req.json().catch(() => ({}));\n const lines: string[] = [];\n if (body.to) lines.push(`To: ${body.to}`);\n if (body.cc) lines.push(`Cc: ${body.cc}`);\n if (body.subject) lines.push(`Subject: ${body.subject}`);\n const raw = encodeRaw(lines, body.body || \"\");\n const res = await fetch(`${BASE}/drafts`, { method: \"POST\", headers, body: JSON.stringify({ message: { raw } }) });\n return json(await safeJson(res), res.status);\n }\n case \"modifyLabels\": {\n const body = await req.json().catch(() => ({}));\n if (!body.id) return json({ error: \"id required\" }, 400);\n const res = await fetch(`${BASE}/messages/${body.id}/modify`, {\n method: \"POST\", headers,\n body: JSON.stringify({ addLabelIds: body.addLabelIds || [], removeLabelIds: body.removeLabelIds || [] }),\n });\n return json(await safeJson(res), res.status);\n }\n\n default:\n if (endpoint === \"message\" && sp.get(\"id\")) {\n url = `${BASE}/messages/${sp.get(\"id\")}?format=full`;\n } else {\n url = `${BASE}/${endpoint}?${params}`;\n }\n }\n\n const res = await fetch(url, { headers });\n let data = await safeJson(res);\n\n // Hydrate list endpoints with full standard headers + labels + snippet.\n const listEndpoints = [\"inbox\", \"messages\", \"unread\", \"sent\", \"starred\"];\n if (listEndpoints.includes(endpoint) && Array.isArray(data?.messages)) {\n const headerQs = STD_HEADERS.map(h => `metadataHeaders=${encodeURIComponent(h)}`).join(\"&\");\n const limited = data.messages.slice(0, HYDRATE_LIMIT);\n const hydrated = await Promise.all(limited.map(async (m: any) => {\n try {\n const r = await fetch(`${BASE}/messages/${m.id}?format=metadata&${headerQs}`, { headers });\n const d = await r.json();\n const hdrs: any[] = d.payload?.headers || [];\n const get = (n: string) => hdrs.find((h) => h.name?.toLowerCase() === n.toLowerCase())?.value ?? \"\";\n return {\n id: m.id, threadId: m.threadId,\n from: get(\"From\"), to: get(\"To\"), cc: get(\"Cc\"),\n subject: get(\"Subject\"), date: get(\"Date\"),\n replyTo: get(\"Reply-To\"), messageId: get(\"Message-ID\"),\n snippet: d.snippet || \"\",\n labelIds: d.labelIds || [],\n sizeEstimate: d.sizeEstimate,\n internalDate: d.internalDate,\n unread: (d.labelIds || []).includes(\"UNREAD\"),\n starred: (d.labelIds || []).includes(\"STARRED\"),\n };\n } catch (e) { return { id: m.id, threadId: m.threadId, error: String(e) }; }\n }));\n data = { ...data, _hydrated: hydrated };\n }\n return json(data, res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(`${BASE}/profile`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: `Gmail unreachable (${res.status})` };\n const p = await safeJson(res);\n return { ok: true, message: `Gmail OK · ${p.emailAddress} · ${p.messagesTotal} messages` };\n}\n",[],[365,917,640],"a3c5be78-5968-11f1-bbfa-ee650bb23af1","c4608676-5806-11f1-9882-ee650bb23af1",310,["D",1779844138416],"// Google Analytics 4 · property-level reports (sessions, users, events, funnels, KPIs).\n// Docs: https://developers.google.com/analytics/devguides/reporting/data/v1/rest\n// Auth: Bearer access_token (shared with central Google integration).\n// All reporting endpoints use POST.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst V1 = `https://analyticsdata.googleapis.com/v1beta/properties`;\nconst V1ALPHA = `https://analyticsdata.googleapis.com/v1alpha/properties`;\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const propertyId = (creds.extra_config?.property_id ?? \"\").toString().trim();\n // QA fix: refuse to build malformed URL like /properties//runReport — used to 400 silently.\n if (!propertyId) {\n return json({ error: \"GA4 property ID required. Add 'Property ID' under Settings → GA4 (e.g. 123456789).\" }, 400);\n }\n if (!creds.access_token) {\n return json({ error: \"GA4 needs Google OAuth. Click 'Connect with Google' in Settings.\" }, 401);\n }\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string, body: string;\n\n switch (endpoint) {\n case \"realtime\":\n url = `${V1}/${propertyId}:runRealtimeReport`;\n body = (await req.text()) || JSON.stringify({\n dimensions: [{ name: \"country\" }],\n metrics: [{ name: \"activeUsers\" }],\n minuteRanges: [{ name: \"last30Minutes\", startMinutesAgo: 29, endMinutesAgo: 0 }],\n });\n break;\n case \"report\": {\n url = `${V1}/${propertyId}:runReport`;\n const reportDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const reportStartDate = `${reportDays}daysAgo`;\n body = (await req.text()) || JSON.stringify({\n dimensions: [{ name: \"sessionDefaultChannelGroup\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"eventCount\" }],\n dateRanges: [{ startDate: reportStartDate, endDate: \"today\" }],\n });\n break;\n }\n case \"user-acquisition\": {\n url = `${V1}/${propertyId}:runReport`;\n const uaDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const uaStartDate = `${uaDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"sessionDefaultChannelGroup\" }],\n metrics: [{ name: \"sessions\" }, { name: \"newUsers\" }, { name: \"engagementRate\" }],\n dateRanges: [{ startDate: uaStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"sessions\" }, desc: true }],\n limit: \"20\",\n });\n break;\n }\n case \"top-events\": {\n url = `${V1}/${propertyId}:runReport`;\n const teDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const teStartDate = `${teDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"eventName\" }],\n metrics: [{ name: \"eventCount\" }, { name: \"totalUsers\" }],\n dateRanges: [{ startDate: teStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"eventCount\" }, desc: true }],\n limit: \"20\",\n });\n break;\n }\n case \"kpi\": {\n url = `${V1}/${propertyId}:runReport`;\n const kpiDays = sp.get(\"days\") || \"30\";\n body = JSON.stringify({\n metrics: [\n { name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"newUsers\" },\n { name: \"engagementRate\" }, { name: \"screenPageViews\" }, { name: \"bounceRate\" },\n ],\n dateRanges: [\n { name: \"current\", startDate: `${kpiDays}daysAgo`, endDate: \"today\" },\n { name: \"previous\", startDate: `${Number(kpiDays) * 2}daysAgo`, endDate: `${Number(kpiDays) + 1}daysAgo` },\n ],\n });\n break;\n }\n case \"traffic-over-time\": {\n url = `${V1}/${propertyId}:runReport`;\n const totDays = sp.get(\"days\") || \"30\";\n body = JSON.stringify({\n dimensions: [{ name: \"date\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }],\n dateRanges: [{ startDate: `${totDays}daysAgo`, endDate: \"today\" }],\n orderBys: [{ dimension: { dimensionName: \"date\" }, desc: false }],\n limit: \"90\",\n });\n break;\n }\n case \"device-breakdown\": {\n url = `${V1}/${propertyId}:runReport`;\n const dbDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const dbStartDate = `${dbDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"deviceCategory\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"bounceRate\" }],\n dateRanges: [{ startDate: dbStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"sessions\" }, desc: true }],\n });\n break;\n }\n case \"page-views\": {\n url = `${V1}/${propertyId}:runReport`;\n const pvDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const pvStartDate = `${pvDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"pagePath\" }],\n metrics: [{ name: \"screenPageViews\" }, { name: \"sessions\" }, { name: \"bounceRate\" }],\n dateRanges: [{ startDate: pvStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"screenPageViews\" }, desc: true }],\n limit: \"25\",\n });\n break;\n }\n case \"funnel\": {\n url = `${V1ALPHA}/${propertyId}:runFunnelReport`;\n const fnDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const fnStartDate = `${fnDays}daysAgo`;\n body = (await req.text()) || JSON.stringify({\n dateRanges: [{ startDate: fnStartDate, endDate: \"today\" }],\n funnel: {\n steps: [\n { name: \"Session Start\", filterExpression: { funnelEventFilter: { eventName: \"session_start\" } } },\n { name: \"Sign Up\", filterExpression: { funnelEventFilter: { eventName: \"sign_up\" } } },\n ],\n },\n });\n break;\n }\n default:\n url = `${V1}/${propertyId}/${endpoint}`;\n body = await req.text();\n }\n\n const res = await fetch(url, { method: \"POST\", headers, body });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,648],"6ddf4bd0-5b48-11f1-bbb6-ee650bb23af1","6ddf4c84-5b48-11f1-bbb6-ee650bb23af1",500,["D",1780050206416],"// Microsoft Excel (Microsoft Graph) · browse workbooks, list worksheets, read cell ranges.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/excel\n// READ: workbooks (list .xlsx) / worksheets (by file id) / range (used range of a sheet)\n// Shares the central Outlook Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me/drive\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n // worksheets/range carry their identifiers either in the query string (direct/AI calls)\n // or in a POST body (the panel renders a form for these). Accept both, query-first.\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n case \"workbooks\": {\n // List Excel workbooks in the user's OneDrive (search by extension).\n url = `${BASE}/root/search(q='.xlsx')?$top=50&$select=id,name,webUrl,lastModifiedDateTime,size`;\n break;\n }\n case \"worksheets\": {\n const id = pick(\"id\");\n if (!id) return json({ error: \"id required (a workbook file id from Workbooks)\" }, 400);\n url = `${BASE}/items/${encodeURIComponent(id)}/workbook/worksheets`;\n break;\n }\n case \"range\": {\n const id = pick(\"id\");\n const sheet = pick(\"sheet\");\n if (!id) return json({ error: \"id required (a workbook file id from Workbooks)\" }, 400);\n if (!sheet) return json({ error: \"sheet required (a worksheet name from Worksheets)\" }, 400);\n url = `${BASE}/items/${encodeURIComponent(id)}/workbook/worksheets/${encodeURIComponent(sheet)}/usedRange?$select=address,values`;\n break;\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,656],"22d42ed0-5d69-11f1-ad33-1607ee4eb77e","c5b2d3da-5806-11f1-8f42-ee650bb23af1",614,["D",1780284156235],"// Google Drive · list/search/create/delete files & folders.\n// Docs: https://developers.google.com/drive/api/reference/rest/v3\n// READ: recent/shared/starred/trash/search/about\n// WRITE: createFolder, deleteFile\n// Returns full file metadata (owners, permissions, sharing, capabilities, version, md5Checksum).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/drive/v3\";\nconst FULL = \"files(id,name,mimeType,description,starred,trashed,parents,modifiedTime,createdTime,viewedByMeTime,size,webViewLink,webContentLink,iconLink,thumbnailLink,owners(displayName,emailAddress,photoLink),lastModifyingUser(displayName,emailAddress,photoLink),sharingUser,shared,permissions(role,type,emailAddress),capabilities,version,md5Checksum,quotaBytesUsed),nextPageToken,incompleteSearch\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"files\":\n case \"recent\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n if (!sp.has(\"orderBy\")) sp.set(\"orderBy\", \"modifiedTime desc\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"shared\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"sharedWithMe = true\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"starred\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"starred = true and trashed = false\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"trash\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"trashed = true\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"search\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n if (!sp.has(\"q\") && sp.has(\"query\")) sp.set(\"q\", `name contains '${sp.get(\"query\")?.replace(/'/g, \"\\\\'\")}'`);\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"about\":\n url = `${BASE}/about?fields=*`;\n break;\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"createFolder\": {\n const body = await req.json().catch(() => ({}));\n if (!body.name) return json({ error: \"name required\" }, 400);\n const res = await fetch(`${BASE}/files`, {\n method: \"POST\", headers,\n body: JSON.stringify({ name: body.name, mimeType: \"application/vnd.google-apps.folder\", parents: body.parents || undefined }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"deleteFile\": {\n const body = await req.json().catch(() => ({}));\n if (!body.id) return json({ error: \"id required\" }, 400);\n const res = await fetch(`${BASE}/files/${body.id}`, { method: \"DELETE\", headers });\n if (res.status === 204) return json({ ok: true, deleted: body.id });\n return json(await safeJson(res), res.status);\n }\n\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,664],"2275de52-5d69-11f1-928d-1607ee4eb77e","c6da0d96-5806-11f1-8602-ee650bb23af1",613,["D",1780284155604],"// Google Docs · list documents, create new, append text.\n// Docs: https://developers.google.com/docs/api/reference/rest\n// READ: list (via Drive), document?id=X\n// WRITE: create, appendText\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst DOCS_BASE = \"https://docs.googleapis.com/v1/documents\";\nconst DRIVE_BASE = \"https://www.googleapis.com/drive/v3/files\";\nconst FULL_FILES = \"files(id,name,mimeType,modifiedTime,createdTime,size,webViewLink,iconLink,owners(displayName,emailAddress),lastModifyingUser(displayName,emailAddress),shared,starred,description),nextPageToken\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"list\": {\n const driveSp = new URLSearchParams({\n q: \"mimeType='application/vnd.google-apps.document' and trashed=false\",\n pageSize: sp.get(\"pageSize\") || \"50\",\n orderBy: \"modifiedTime desc\",\n fields: FULL_FILES,\n });\n url = `${DRIVE_BASE}?${driveSp.toString()}`;\n break;\n }\n case \"document\": {\n const id = sp.get(\"documentId\");\n if (!id) return json({ error: \"documentId param required\" }, 400);\n url = `${DOCS_BASE}/${id}`;\n break;\n }\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"create\": {\n const body = await req.json().catch(() => ({}));\n if (!body.title) return json({ error: \"title required\" }, 400);\n const createRes = await fetch(DOCS_BASE, {\n method: \"POST\", headers, body: JSON.stringify({ title: body.title }),\n });\n const doc = await safeJson(createRes);\n if (createRes.ok && body.content && doc.documentId) {\n await fetch(`${DOCS_BASE}/${doc.documentId}:batchUpdate`, {\n method: \"POST\", headers,\n body: JSON.stringify({ requests: [{ insertText: { location: { index: 1 }, text: body.content } }] }),\n });\n }\n return json(doc, createRes.status);\n }\n case \"appendText\": {\n const body = await req.json().catch(() => ({}));\n if (!body.documentId || !body.text) return json({ error: \"documentId + text required\" }, 400);\n const docRes = await fetch(`${DOCS_BASE}/${body.documentId}?fields=body.content`, { headers });\n const docData = await safeJson(docRes);\n const endIndex = docData?.body?.content?.slice(-1)?.[0]?.endIndex || 1;\n const res = await fetch(`${DOCS_BASE}/${body.documentId}:batchUpdate`, {\n method: \"POST\", headers,\n body: JSON.stringify({ requests: [{ insertText: { location: { index: endIndex - 1 }, text: body.text } }] }),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${DOCS_BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,672],"c73dcd86-5806-11f1-ada3-ee650bb23af1","c73dce58-5806-11f1-ada3-ee650bb23af1",119,["D",1779692156085],"// Google Calendar · list/create/delete events.\n// Docs: https://developers.google.com/calendar/api/v3/reference\n// READ: events/upcoming/today/week/calendars/settings\n// WRITE: createEvent, deleteEvent\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/calendar/v3\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"events\":\n case \"upcoming\":\n if (!sp.has(\"timeMin\")) sp.set(\"timeMin\", new Date().toISOString());\n if (!sp.has(\"timeMax\")) sp.set(\"timeMax\", new Date(Date.now() + 30 * 86400000).toISOString());\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n case \"calendars\":\n url = `${BASE}/users/me/calendarList?maxResults=250`;\n break;\n case \"today\": {\n const start = new Date(); start.setHours(0,0,0,0);\n const end = new Date(); end.setHours(23,59,59,999);\n sp.set(\"timeMin\", start.toISOString());\n sp.set(\"timeMax\", end.toISOString());\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n sp.set(\"maxResults\", \"50\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n }\n case \"week\": {\n const start = new Date(); start.setHours(0,0,0,0);\n const end = new Date(Date.now() + 7 * 86400000);\n sp.set(\"timeMin\", start.toISOString());\n sp.set(\"timeMax\", end.toISOString());\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n sp.set(\"maxResults\", \"100\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n }\n case \"settings\":\n url = `${BASE}/users/me/settings`;\n break;\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"createEvent\": {\n const body = await req.json().catch(() => ({}));\n if (!body.summary || !body.start || !body.end) return json({ error: \"summary, start, end required\" }, 400);\n const calId = body.calendarId || \"primary\";\n const payload: any = {\n summary: body.summary,\n description: body.description,\n location: body.location,\n start: { dateTime: new Date(body.start).toISOString() },\n end: { dateTime: new Date(body.end).toISOString() },\n };\n if (Array.isArray(body.attendees)) payload.attendees = body.attendees.map((email: string) => ({ email }));\n const res = await fetch(`${BASE}/calendars/${calId}/events?sendUpdates=all`, {\n method: \"POST\", headers, body: JSON.stringify(payload),\n });\n return json(await safeJson(res), res.status);\n }\n case \"deleteEvent\": {\n const body = await req.json().catch(() => ({}));\n if (!body.eventId) return json({ error: \"eventId required\" }, 400);\n const calId = body.calendarId || \"primary\";\n const res = await fetch(`${BASE}/calendars/${calId}/events/${body.eventId}`, { method: \"DELETE\", headers });\n if (res.status === 204) return json({ ok: true, deleted: body.eventId });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,680],"c3d36e94-5806-11f1-8bf1-ee650bb23af1","c3d36f5c-5806-11f1-8bf1-ee650bb23af1",112,["D",1779692150355],"// Ahrefs · SEO data (keywords, backlinks, top pages, domain rating, organic traffic).\n// Docs: https://docs.ahrefs.com/en/api/reference/site-explorer\n// Auth: Bearer api_key. Important quirks:\n// - /metrics, /domain-rating, /backlinks-stats do NOT accept a `select` parameter.\n// - /all-backlinks, /organic-keywords, /keywords-explorer/matching-terms REQUIRE `select`.\n// - The referring-domains endpoint path is /refdomains (not /referring-domains).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://api.ahrefs.com/v3\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.api_key}` };\n const today = new Date().toISOString().slice(0, 10);\n const sp = new URLSearchParams(params);\n const target = sp.get(\"target\") || creds.extra_config?.target || \"\";\n if (target) sp.set(\"target\", target);\n\n let url: string;\n switch (endpoint) {\n case \"site-explorer\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/metrics?${sp.toString()}`;\n break;\n }\n case \"keywords\": {\n if (!sp.has(\"select\")) sp.set(\"select\", \"keyword,volume,difficulty,cpc,traffic_potential,global_volume\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n const seed = sp.get(\"keywords\") || sp.get(\"keyword\") || sp.get(\"q\") || \"seo\";\n sp.delete(\"keyword\"); sp.delete(\"q\");\n sp.set(\"keywords\", seed);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/keywords-explorer/matching-terms?${sp.toString()}`;\n break;\n }\n case \"backlinks\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"url_from,domain_rating_source,anchor,name_source,title,first_seen,is_dofollow,traffic\");\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/all-backlinks?${sp.toString()}`;\n break;\n }\n case \"domain-rating\": {\n if (!sp.get(\"target\")) sp.set(\"target\", \"ahrefs.com\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/domain-rating?${sp.toString()}`;\n break;\n }\n case \"organic-keywords\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"keyword,volume,sum_traffic,best_position,cpc,keyword_difficulty\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/organic-keywords?${sp.toString()}`;\n break;\n }\n case \"top-pages\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"url,sum_traffic,keywords,top_keyword,top_keyword_best_position,value\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/top-pages?${sp.toString()}`;\n break;\n }\n case \"referring-domains\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"domain,domain_rating,dofollow_links,dofollow_linked_domains,traffic_domain,first_seen\");\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/refdomains?${sp.toString()}`;\n break;\n }\n case \"backlinks-stats\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/backlinks-stats?${sp.toString()}`;\n break;\n }\n default:\n return json({ error: `Unknown Ahrefs endpoint: ${endpoint}` }, 400);\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const today = new Date().toISOString().slice(0, 10);\n const res = await fetch(`${BASE}/site-explorer/domain-rating?target=ahrefs.com&date=${today}`, {\n headers: { Authorization: `Bearer ${creds.api_key}` },\n });\n if (res.status === 401 || res.status === 403) return { ok: false, message: \"Invalid API key\" };\n if (res.status === 429) return { ok: false, message: \"Rate limited — try again shortly\" };\n if (!res.ok) {\n const d = await safeJson(res);\n return { ok: false, message: d?.error ?? `API error ${res.status}` };\n }\n return { ok: true, message: \"API key verified\" };\n}\n",[],[365,917,688],"212361f0-5d69-11f1-bb77-1607ee4eb77e","_shared.ts","c3708540-5806-11f1-acd7-ee650bb23af1",612,["D",1780284153389],"// Common utilities shared across per-integration handler files.\n// Each integration handler file imports from here rather than duplicating these helpers.\nimport { sqlite } from \"../../db.ts\";\n\nexport interface IntegrationCredentials {\n api_key?: string;\n api_secret?: string;\n access_token?: string;\n refresh_token?: string;\n extra_config?: Record<string, any>;\n}\n\nconst CORS = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, PUT, DELETE, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n};\n\nexport function json(data: any, status = 200): Response {\n return new Response(JSON.stringify(data), {\n status,\n headers: { \"Content-Type\": \"application/json\", ...CORS },\n });\n}\n\nexport async function safeJson(res: Response): Promise<any> {\n const text = await res.text();\n try { return JSON.parse(text); } catch { return { error: text.slice(0, 500) || `HTTP ${res.status}` }; }\n}\n\n// SSRF guard for any server-side fetch built from user-supplied URLs (WordPress site_url,\n// custom LLM endpoint). Only allows public https hosts; blocks loopback/private/link-local\n// ranges and the cloud metadata IP. (Defense in depth — the APP_TOKEN gate already restricts\n// who can configure these.)\nexport function isSafePublicUrl(rawUrl: string): boolean {\n let u: URL;\n try { u = new URL(rawUrl); } catch { return false; }\n if (u.protocol !== \"https:\") return false;\n const host = u.hostname.toLowerCase().replace(/^\\[|\\]$/g, \"\");\n if (host === \"localhost\" || host.endsWith(\".localhost\") || host.endsWith(\".internal\") || host.endsWith(\".local\")) return false;\n if (host === \"::1\" || host === \"::\" || host.startsWith(\"fe80:\") || host.startsWith(\"fc\") || host.startsWith(\"fd\")) return false;\n const m = host.match(/^(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})$/);\n if (m) {\n const a = +m[1], b = +m[2];\n if (a === 0 || a === 127) return false; // unspecified / loopback\n if (a === 10) return false; // 10/8 private\n if (a === 172 && b >= 16 && b <= 31) return false; // 172.16/12 private\n if (a === 192 && b === 168) return false; // 192.168/16 private\n if (a === 169 && b === 254) return false; // link-local / cloud metadata\n }\n return true;\n}\n\nexport async function getCredentials(name: string): Promise<IntegrationCredentials | null> {\n const result = await sqlite.execute(\n \"SELECT api_key, api_secret, access_token, refresh_token, extra_config FROM integrations WHERE name = ?\",\n [name]\n );\n if (!result.rows || result.rows.length === 0) {\n // Google-family services (drive/sheets/docs/gmail/etc.) share the central 'google' OAuth\n // credentials and usually have NO row of their own. Inherit the google creds so the router\n // doesn't 404 before withGoogleFallback runs. ('google' is not in GOOGLE_SERVICES → no recursion.)\n if (GOOGLE_SERVICES.includes(name)) return await getCredentials(\"google\");\n // Microsoft-family services (onedrive/teams/excel) share the central 'outlook' Graph token.\n if (MICROSOFT_SERVICES.includes(name)) return await getCredentials(\"outlook\");\n return null;\n }\n const cols = result.columns;\n const row = result.rows[0];\n const obj: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) obj[cols[i]] = row[i];\n let extra_config: Record<string, any> | undefined;\n if (obj.extra_config) {\n try { extra_config = JSON.parse(obj.extra_config); } catch { extra_config = {}; }\n }\n return {\n api_key: obj.api_key ?? undefined,\n api_secret: obj.api_secret ?? undefined,\n access_token: obj.access_token ?? undefined,\n refresh_token: obj.refresh_token ?? undefined,\n extra_config,\n };\n}\n\n// All Google services share the OAuth token issued for `google`.\nexport async function withGoogleFallback(creds: IntegrationCredentials): Promise<IntegrationCredentials> {\n const google = await getCredentials(\"google\");\n return {\n ...creds,\n access_token: creds.access_token || google?.access_token,\n refresh_token: creds.refresh_token || google?.refresh_token,\n api_key: creds.api_key || google?.api_key,\n extra_config: { ...google?.extra_config, ...creds.extra_config },\n };\n}\n\n// Refresh Google access_token using stored refresh_token. Persists the new token and returns it.\nexport async function refreshGoogleToken(): Promise<string | null> {\n const google = await getCredentials(\"google\");\n if (!google?.refresh_token) return null;\n // Match the callback handler: env-driven OAuth client by design, DB row as fallback. Trim to\n // avoid whitespace producing a malformed (invalid_request) token request.\n const clientId = (Deno.env.get(\"GOOGLE_CLIENT_ID\") || google.extra_config?.client_id || google.extra_config?.clientId || \"\").trim();\n const clientSecret = (Deno.env.get(\"GOOGLE_CLIENT_SECRET\") || google.api_secret || \"\").trim();\n if (!clientId || !clientSecret) return null;\n const res = await fetch(\"https://oauth2.googleapis.com/token\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({ refresh_token: google.refresh_token, client_id: clientId, client_secret: clientSecret, grant_type: \"refresh_token\" }),\n });\n const data = await res.json();\n if (!data.access_token) return null;\n await sqlite.execute(\"UPDATE integrations SET access_token=?, updated_at=? WHERE name='google'\", [data.access_token, new Date().toISOString()]);\n return data.access_token;\n}\n\n// List of integrations that share the central Google OAuth credentials.\nexport const GOOGLE_SERVICES = [\"ga4\", \"gsc\", \"youtube\", \"drive\", \"sheets\", \"calendar\", \"gmail\", \"docs\"];\n\n// Env-driven Google OAuth client helpers. The OAuth client id/secret live in the\n// GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET env vars by design. Re-exported by api/integrations.ts;\n// removing these breaks that module's import graph (and the whole app).\nexport const googleClientId = (): string => (Deno.env.get(\"GOOGLE_CLIENT_ID\") || \"\").trim();\nexport const googleClientSecret = (): string => (Deno.env.get(\"GOOGLE_CLIENT_SECRET\") || \"\").trim();\nexport const googleRedirectUri = (): string => (Deno.env.get(\"GOOGLE_REDIRECT_URI\") || \"\").trim();\n\n// ─── Microsoft (Outlook / Graph) OAuth ──────────────────────────────────────\n// Env-driven confidential client, same pattern as Google. `common` tenant accepts both\n// work/school and personal Microsoft accounts. Delegated scopes cover the Outlook app's\n// read (inbox/calendar/contacts) + write (send) endpoints; offline_access yields a refresh token.\nexport const MICROSOFT_TENANT = \"common\";\nexport const MICROSOFT_SCOPES =\n \"openid profile email offline_access User.Read Mail.Read Mail.Send Calendars.Read Contacts.Read \" +\n \"Files.Read.All Files.ReadWrite.All Team.ReadBasic.All Channel.ReadBasic.All ChannelMessage.Read.All \" +\n \"ChannelMessage.Send Chat.ReadWrite ChatMessage.Send OnlineMeetings.ReadWrite\";\nexport const microsoftClientId = (): string => (Deno.env.get(\"MICROSOFT_CLIENT_ID\") || \"\").trim();\nexport const microsoftClientSecret = (): string => (Deno.env.get(\"MICROSOFT_CLIENT_SECRET\") || \"\").trim();\nexport const microsoftTokenUrl = (): string =>\n `https://login.microsoftonline.com/${MICROSOFT_TENANT}/oauth2/v2.0/token`;\n\n// Refresh the Outlook access_token using its stored refresh_token. Persists the new token\n// (and the rotated refresh_token, if Microsoft returns one) and returns the new access token.\nexport async function refreshMicrosoftToken(): Promise<string | null> {\n const outlook = await getCredentials(\"outlook\");\n if (!outlook?.refresh_token) return null;\n // Env-first, falling back to the stored outlook credential row (same as the OAuth routes).\n const clientId = (microsoftClientId() || outlook.extra_config?.client_id || outlook.extra_config?.clientId || \"\").trim();\n const clientSecret = (microsoftClientSecret() || outlook.api_secret || \"\").trim();\n if (!clientId || !clientSecret) return null;\n const res = await fetch(microsoftTokenUrl(), {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({\n refresh_token: outlook.refresh_token,\n client_id: clientId,\n client_secret: clientSecret,\n grant_type: \"refresh_token\",\n scope: MICROSOFT_SCOPES,\n }),\n });\n const data = await res.json();\n if (!data.access_token) return null;\n await sqlite.execute(\n \"UPDATE integrations SET access_token=?, refresh_token=COALESCE(?, refresh_token), updated_at=? WHERE name='outlook'\",\n [data.access_token, data.refresh_token ?? null, new Date().toISOString()],\n );\n return data.access_token;\n}\n\n// Microsoft Graph services that share the central Outlook OAuth token (mirror GOOGLE_SERVICES).\nexport const MICROSOFT_SERVICES = [\"onedrive\", \"teams\", \"excel\"];\n\n// All Microsoft services share the OAuth token issued for `outlook` (the Graph token).\nexport async function withMicrosoftFallback(creds: IntegrationCredentials): Promise<IntegrationCredentials> {\n const outlook = await getCredentials(\"outlook\");\n return {\n ...creds,\n access_token: creds.access_token || outlook?.access_token,\n refresh_token: creds.refresh_token || outlook?.refresh_token,\n api_secret: creds.api_secret || outlook?.api_secret,\n extra_config: { ...outlook?.extra_config, ...creds.extra_config },\n };\n}\n",[],[365,917,1025],"c370b894-5806-11f1-acd7-ee650bb23af1",111,["D",1779692149653],[382,383,384,385,386,387,388,389,390,391,392,393,394,395,396,397,398],[365,917],["D",1779556136787],[381,399,400,401,402,403,404,405,406,407,408,409,410,411,412,413,414,415,416,417,418,419,420,421,422,423],[365],["D",1779556068488],["D",1789503703835],{"_99":98,"_106":102},{"_99":98,"_100":98,"_101":102,"_18":54,"_103":104,"_105":24},{"_13":88,"_31":89,"_42":41,"_90":43,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":44,"_97":98},"ctx",{"_81":-5,"_126":1078,"_1079":1080,"_1081":1082,"_1083":24,"_1084":24,"_1085":24,"_1086":58,"_1087":-5,"_1088":-7,"_1089":-7,"_1090":62,"_1091":24,"_1092":24,"_1093":-5,"_1094":24,"_1095":24,"_1096":24,"_1097":-5},"ENV",{"_1049":1050,"_1051":1052,"_1053":89,"_1054":1055,"_1056":1057,"_1058":1059,"_1060":1061,"_1062":1063,"_1064":1065,"_1066":1067,"_1068":1069,"_1070":1071,"_1072":1073,"_1074":1075,"_1076":1077},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"] </article></body></html>

[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_1045":1046,"_1047":1048},"routes/_app.x.$handle.$name",{"_99":98,"_31":44,"_107":89,"_26":108,"_13":14,"_109":-5,"_11":110,"_111":112,"_113":114,"_115":116,"_117":118,"_20":-5,"_119":-5,"_120":-5,"_121":24,"_122":123,"_124":58,"_125":-5,"_126":-5,"_127":128},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_42":43,"_31":44,"_37":45,"_46":-5,"_47":24,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":66,"_67":68,"_69":70,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":82,"_83":84,"_85":86,"_87":24},"version",645,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":-7,"_23":24,"_25":24,"_26":27,"_28":27},"file",{"_29":30,"_31":32,"_33":34,"_35":36,"_37":38,"_39":14,"_40":41,"_23":24},"previous","deleted",false,"deletedHere","path","skills/catalog/part_2.ts","latestPath","fileId","ec2a718c-5d6d-11f1-9fc6-1607ee4eb77e","name","part_2.ts","fileType","script","content","// GENERATED by deploy/gen_skills_catalog.mjs — do not edit by hand.\nimport type { SkillDef } from \"../index.ts\";\nexport const SKILLS_PART_2: SkillDef[] = [\n {\"name\":\"analytics\",\"category\":\"Analytics\",\"description\":\"When the user wants to set up, improve, or audit analytics tracking and measurement. Also use when the user mentions \\\"set up tracking,\\\" \\\"GA4,\\\" \\\"Google Analytics,\\\" \\\"conversion tracking,\\\" \\\"event tracking,\\\" \\\"UTM parameters,\\\" \\\"tag manager,\\\" \\\"GTM,\\\" \\\"analytics implementation,\\\" \\\"tracking plan,\\\" \\\"how do I measure this,\\\" \\\"track conversions,\\\" \\\"attribution,\\\" \\\"Mixpanel,\\\" \\\"Segment,\\\" \\\"are my events firing,\\\" or \\\"analytics isn't working.\\\" Use this whenever someone asks how to know if something is working or wants to measure marketing results. For A/B test measurement, see ab-testing.\",\"version\":\"2.0.0\",\"body\":\"# Analytics Tracking\\n\\nYou are an expert in analytics implementation and measurement. Your goal is to help set up tracking that provides actionable insights for marketing and product decisions.\\n\\n## Initial Assessment\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nBefore implementing tracking, understand:\\n\\n1. **Business Context** - What decisions will this data inform? What are key conversions?\\n2. **Current State** - What tracking exists? What tools are in use?\\n3. **Technical Context** - What's the tech stack? Any privacy/compliance requirements?\\n\\n---\\n\\n## Core Principles\\n\\n### 1. Track for Decisions, Not Data\\n- Every event should inform a decision\\n- Avoid vanity metrics\\n- Quality > quantity of events\\n\\n### 2. Start with the Questions\\n- What do you need to know?\\n- What actions will you take based on this data?\\n- Work backwards to what you need to track\\n\\n### 3. Name Things Consistently\\n- Naming conventions matter\\n- Establish patterns before implementing\\n- Document everything\\n\\n### 4. Maintain Data Quality\\n- Validate implementation\\n- Monitor for issues\\n- Clean data > more data\\n\\n---\\n\\n## Tracking Plan Framework\\n\\n### Structure\\n\\n```\\nEvent Name | Category | Properties | Trigger | Notes\\n---------- | -------- | ---------- | ------- | -----\\n```\\n\\n### Event Types\\n\\n| Type | Examples |\\n|------|----------|\\n| Pageviews | Automatic, enhanced with metadata |\\n| User Actions | Button clicks, form submissions, feature usage |\\n| System Events | Signup completed, purchase, subscription changed |\\n| Custom Conversions | Goal completions, funnel stages |\\n\\n**For comprehensive event lists**: See [references/event-library.md](references/event-library.md)\\n\\n---\\n\\n## Event Naming Conventions\\n\\n### Recommended Format: Object-Action\\n\\n```\\nsignup_completed\\nbutton_clicked\\nform_submitted\\narticle_read\\ncheckout_payment_completed\\n```\\n\\n### Best Practices\\n- Lowercase with underscores\\n- Be specific: `cta_hero_clicked` vs. `button_clicked`\\n- Include context in properties, not event name\\n- Avoid spaces and special characters\\n- Document decisions\\n\\n---\\n\\n## Essential Events\\n\\n### Marketing Site\\n\\n| Event | Properties |\\n|-------|------------|\\n| cta_clicked | button_text, location |\\n| form_submitted | form_type |\\n| signup_completed | method, source |\\n| demo_requested | - |\\n\\n### Product/App\\n\\n| Event | Properties |\\n|-------|------------|\\n| onboarding_step_completed | step_number, step_name |\\n| feature_used | feature_name |\\n| purchase_completed | plan, value |\\n| subscription_cancelled | reason |\\n\\n**For full event library by business type**: See [references/event-library.md](references/event-library.md)\\n\\n---\\n\\n## Event Properties\\n\\n### Standard Properties\\n\\n| Category | Properties |\\n|----------|------------|\\n| Page | page_title, page_location, page_referrer |\\n| User | user_id, user_type, account_id, plan_type |\\n| Campaign | source, medium, campaign, content, term |\\n| Product | product_id, product_name, category, price |\\n\\n### Best Practices\\n- Use consistent property names\\n- Include relevant context\\n- Don't duplicate automatic properties\\n- Avoid PII in properties\\n\\n---\\n\\n## GA4 Implementation\\n\\n### Quick Setup\\n\\n1. Create GA4 property and data stream\\n2. Install gtag.js or GTM\\n3. Enable enhanced measurement\\n4. Configure custom events\\n5. Mark conversions in Admin\\n\\n### Custom Event Example\\n\\n```javascript\\ngtag('event', 'signup_completed', {\\n 'method': 'email',\\n 'plan': 'free'\\n});\\n```\\n\\n**For detailed GA4 implementation**: See [references/ga4-implementation.md](references/ga4-implementation.md)\\n\\n---\\n\\n## Google Tag Manager\\n\\n### Container Structure\\n\\n| Component | Purpose |\\n|-----------|---------|\\n| Tags | Code that executes (GA4, pixels) |\\n| Triggers | When tags fire (page view, click) |\\n| Variables | Dynamic values (click text, data layer) |\\n\\n### Data Layer Pattern\\n\\n```javascript\\ndataLayer.push({\\n 'event': 'form_submitted',\\n 'form_name': 'contact',\\n 'form_location': 'footer'\\n});\\n```\\n\\n**For detailed GTM implementation**: See [references/gtm-implementation.md](references/gtm-implementation.md)\\n\\n---\\n\\n## UTM Parameter Strategy\\n\\n### Standard Parameters\\n\\n| Parameter | Purpose | Example |\\n|-----------|---------|---------|\\n| utm_source | Traffic source | google, newsletter |\\n| utm_medium | Marketing medium | cpc, email, social |\\n| utm_campaign | Campaign name | spring_sale |\\n| utm_content | Differentiate versions | hero_cta |\\n| utm_term | Paid search keywords | running+shoes |\\n\\n### Naming Conventions\\n- Lowercase everything\\n- Use underscores or hyphens consistently\\n- Be specific but concise: `blog_footer_cta`, not `cta1`\\n- Document all UTMs in a spreadsheet\\n\\n---\\n\\n## Debugging and Validation\\n\\n### Testing Tools\\n\\n| Tool | Use For |\\n|------|---------|\\n| GA4 DebugView | Real-time event monitoring |\\n| GTM Preview Mode | Test triggers before publish |\\n| Browser Extensions | Tag Assistant, dataLayer Inspector |\\n\\n### Validation Checklist\\n\\n- [ ] Events firing on correct triggers\\n- [ ] Property values populating correctly\\n- [ ] No duplicate events\\n- [ ] Works across browsers and mobile\\n- [ ] Conversions recorded correctly\\n- [ ] No PII leaking\\n\\n### Common Issues\\n\\n| Issue | Check |\\n|-------|-------|\\n| Events not firing | Trigger config, GTM loaded |\\n| Wrong values | Variable path, data layer structure |\\n| Duplicate events | Multiple containers, trigger firing twice |\\n\\n---\\n\\n## Privacy and Compliance\\n\\n### Considerations\\n- Cookie consent required in EU/UK/CA\\n- No PII in analytics properties\\n- Data retention settings\\n- User deletion capabilities\\n\\n### Implementation\\n- Use consent mode (wait for consent)\\n- IP anonymization\\n- Only collect what you need\\n- Integrate with consent management platform\\n\\n---\\n\\n## Output Format\\n\\n### Tracking Plan Document\\n\\n```markdown\\n# [Site/Product] Tracking Plan\\n\\n## Overview\\n- Tools: GA4, GTM\\n- Last updated: [Date]\\n\\n## Events\\n\\n| Event Name | Description | Properties | Trigger |\\n|------------|-------------|------------|---------|\\n| signup_completed | User completes signup | method, plan | Success page |\\n\\n## Custom Dimensions\\n\\n| Name | Scope | Parameter |\\n|------|-------|-----------|\\n| user_type | User | user_type |\\n\\n## Conversions\\n\\n| Conversion | Event | Counting |\\n|------------|-------|----------|\\n| Signup | signup_completed | Once per session |\\n```\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What tools are you using (GA4, Mixpanel, etc.)?\\n2. What key actions do you want to track?\\n3. What decisions will this data inform?\\n4. Who implements - dev team or marketing?\\n5. Are there privacy/consent requirements?\\n6. What's already tracked?\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md). Key analytics tools:\\n\\n| Tool | Best For | MCP | Guide |\\n|------|----------|:---:|-------|\\n| **GA4** | Web analytics, Google ecosystem | ✓ | [ga4.md](../../tools/integrations/ga4.md) |\\n| **Mixpanel** | Product analytics, event tracking | - | [mixpanel.md](../../tools/integrations/mixpanel.md) |\\n| **Amplitude** | Product analytics, cohort analysis | - | [amplitude.md](../../tools/integrations/amplitude.md) |\\n| **PostHog** | Open-source analytics, session replay | - | [posthog.md](../../tools/integrations/posthog.md) |\\n| **Segment** | Customer data platform, routing | - | [segment.md](../../tools/integrations/segment.md) |\\n\\n---\\n\\n## Related Skills\\n\\n- **ab-testing**: For experiment tracking\\n- **seo-audit**: For organic traffic analysis\\n- **cro**: For conversion optimization (uses this data)\\n- **revops**: For pipeline metrics, CRM tracking, and revenue attribution\"},\n {\"name\":\"aso\",\"category\":\"SEO\",\"description\":\"\\\"When the user wants to audit or optimize an App Store or Google Play listing. Also use when the user mentions 'ASO audit,' 'app store optimization,' 'optimize my app listing,' 'improve app visibility,' 'app store ranking,' 'audit my listing,' 'why aren't people downloading my app,' 'improve my app conversion,' 'keyword optimization for app,' or 'compare my app to competitors.' Use when the user shares an App Store or Google Play URL and wants to improve it.\\\"\",\"version\":\"2.0.0\",\"body\":\"# ASO Audit\\n\\nAnalyze App Store and Google Play listings against ASO best practices. Fetches\\nlive listing data, scores metadata, visuals, and ratings, then produces a\\nprioritized action plan.\\n\\n## When to Use\\n\\n- User shares an App Store or Google Play URL\\n- User asks to audit or optimize an app listing\\n- User wants to compare their app against competitors\\n- User asks about app store ranking, visibility, or download conversion\\n\\n## Before Auditing\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\n## Phase 1 — Identify Store & Fetch\\n\\n### Detect store type from URL\\n\\n```\\nApple: apps.apple.com/{country}/app/{name}/id{digits}\\nGoogle: play.google.com/store/apps/details?id={package}\\n```\\n\\nIf the user gives an app name instead of a URL, search the web for:\\n`site:apps.apple.com \\\"{app name}\\\"` or `site:play.google.com \\\"{app name}\\\"`\\n\\n### Fetch the listing\\n\\nUse WebFetch to retrieve the listing page. Extract every available field:\\n\\n**Apple App Store fields:**\\n\\n- App name (title) — 30 char limit\\n- Subtitle — 30 char limit\\n- Description (long) — not indexed for search, but matters for conversion\\n- Promotional text — 170 chars, updatable without new release\\n- Category (primary + secondary)\\n- Screenshots (count, order, caption text)\\n- Preview video (presence, duration)\\n- Rating (average + count)\\n- Recent reviews (visible ones)\\n- Price / in-app purchases\\n- Developer name\\n- Last updated date\\n- Version history notes\\n- Age rating\\n- Size\\n- Languages / localizations listed\\n- In-app events (if any visible)\\n\\n**Google Play fields:**\\n\\n- App name (title) — 30 char limit\\n- Short description — 80 char limit\\n- Full description — 4,000 char limit, IS indexed for search\\n- Category + tags\\n- Feature graphic (presence)\\n- Screenshots (count, order)\\n- Preview video (presence)\\n- Rating (average + count)\\n- Recent reviews (visible ones)\\n- Price / in-app purchases\\n- Developer name\\n- Last updated date\\n- What's new text\\n- Downloads range\\n- Content rating\\n- Data safety section\\n- Languages listed\\n\\nIf WebFetch returns incomplete data (stores render client-side), note gaps and\\nwork with what's available. Ask the user to paste missing fields if critical.\\n\\n### Visual asset assessment\\n\\nWebFetch cannot extract screenshot images or caption text. **Take a screenshot\\nof the listing page** to get visual data:\\n\\n1. Navigate to the listing URL and capture a full-page screenshot\\n2. Assess the screenshot for: icon quality, screenshot count, caption text,\\n messaging quality, preview video presence, feature graphic (Google Play)\\n3. If browser tools are unavailable, ask the user to share a screenshot of the\\n listing page\\n\\n**Promotional text (Apple):** This 170-char field appears above the description\\nbut is often indistinguishable from it in scraped HTML. If you cannot confirm\\nits presence, note this and recommend the user check App Store Connect.\\n\\n---\\n\\n## Phase 1.5 — Assess Brand Maturity\\n\\nBefore scoring, classify the app into one of three tiers. This determines how\\nyou interpret \\\"textbook ASO\\\" deviations — a deliberate brand choice by a\\nhousehold name is not the same as a missed opportunity by an unknown app.\\n\\n### Tier definitions\\n\\n| Tier | Signals | Examples |\\n| --------------- | ------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------- |\\n| **Dominant** | Household name, 1M+ ratings, top-10 in category, near-universal brand recognition. Users search by brand name, not generic keywords. | Instagram, Uber, Spotify, WhatsApp, Netflix |\\n| **Established** | Well-known in their category, 100K+ ratings, strong organic installs, recognized brand but not universally known. | Strava, Notion, Duolingo, Cash App, Calm |\\n| **Challenger** | Building awareness, <100K ratings, needs discovery through keywords and ASO tactics. Most apps fall here. | Your app, most indie/startup apps |\\n\\n### How tier affects scoring\\n\\n**Dominant apps** get adjusted scoring in these areas:\\n\\n- **Title:** Brand-only or brand-first titles are valid (score 8+ if brand is the keyword). These apps don't need generic keyword discovery.\\n- **Description:** Score purely on conversion quality, not keyword presence. If the app is a household name, a well-crafted brand description beats a keyword-stuffed one.\\n- **Visual Assets:** Lifestyle/brand photography instead of UI demos is a legitimate conversion strategy. No video is acceptable if the product is hard to demo in 30s or brand awareness is near-universal.\\n- **What's New:** Generic release notes at weekly+ cadence are acceptable (score 8+). At scale, detailed changelogs have minimal ROI and risk backlash.\\n- **In-app events:** Missing events for utility apps with massive install bases (Uber, WhatsApp) is not a penalty. These apps don't need discovery help.\\n- **Localization:** Score relative to actual market, not absolute count. A US-only fintech with 2 languages (English + Spanish) is appropriately localized.\\n\\n**Established apps** get partial adjustment:\\n\\n- Brand-first titles are fine but should still include 1-2 keywords\\n- Strategic description choices get benefit of the doubt\\n- Other dimensions scored normally\\n\\n**Challenger apps** are scored strictly against textbook ASO best practices — every character, screenshot, and keyword matters.\\n\\n**Key principle:** Before docking points, ask: \\\"Is this a mistake or a deliberate\\nchoice by a team that has data I don't?\\\" If the app has 1M+ ratings and a\\ndedicated ASO team, assume their choices are data-informed unless clearly wrong.\\n\\n---\\n\\n## Phase 2 — Score Each Dimension\\n\\nScore each dimension 0-10 using the criteria in `references/scoring-criteria.md`.\\nApply the brand maturity tier adjustments from Phase 1.5.\\n\\nReference files for platform specs and benchmarks:\\n\\n- `references/apple-specs.md` — Official Apple character limits, screenshot/video specs, CPP/PPO rules, rejection triggers\\n- `references/google-play-specs.md` — Official Google Play limits, screenshot specs, Android Vitals thresholds, policies\\n- `references/benchmarks.md` — Conversion data, rating impact, video lift, screenshot behavior, CPP/event benchmarks\\n\\n### Dimensions and Weights\\n\\n| # | Dimension | Weight | What It Covers |\\n| --- | -------------------- | ------ | ------------------------------------------------------------------------- |\\n| 1 | Title & Subtitle | 20% | Character usage, keyword presence, clarity, brand + keyword balance |\\n| 2 | Description | 15% | First 3 lines, keyword density (Google), CTA, structure, promotional text |\\n| 3 | Visual Assets | 25% | Screenshot count/quality/messaging, video, icon, feature graphic |\\n| 4 | Ratings & Reviews | 20% | Average rating, volume, recency, developer responses |\\n| 5 | Metadata & Freshness | 10% | Category choice, update recency, localization count, data safety |\\n| 6 | Conversion Signals | 10% | Price positioning, IAP transparency, social proof, download range |\\n\\n**Final score** = weighted sum, out of 100.\\n\\n### Score interpretation\\n\\n| Score | Grade | Meaning |\\n| ------ | ----- | --------------------------------------------------------- |\\n| 85-100 | A | Well-optimized; focus on A/B testing and iteration |\\n| 70-84 | B | Good foundation; clear opportunities to improve |\\n| 50-69 | C | Significant gaps; prioritized fixes will have high impact |\\n| 30-49 | D | Major optimization needed across multiple dimensions |\\n| 0-29 | F | Listing needs a complete overhaul |\\n\\n---\\n\\n## Phase 3 — Competitor Comparison (Optional)\\n\\nIf the user provides competitor URLs or asks for comparison:\\n\\n1. Fetch 2-3 top competitors in the same category\\n2. Run the same scoring on each\\n3. Build a comparison table highlighting where the user's app is weaker/stronger\\n4. Identify keyword gaps — terms competitors rank for that the user's app doesn't target\\n\\nIf no competitors are specified, suggest the user provide 2-3 or offer to search\\nfor top apps in their category.\\n\\n---\\n\\n## Phase 4 — Generate Report\\n\\nUse the template in `references/report-template.md` to structure the output.\\n\\nThe report must include:\\n\\n1. **Score card** — table with all 6 dimensions, scores, and grade\\n2. **Top 3 quick wins** — changes that take <1 hour and have highest impact\\n3. **Detailed findings** — per-dimension breakdown with specific issues and fixes\\n4. **Keyword suggestions** — based on title/description analysis and competitor gaps\\n5. **Visual asset recommendations** — specific screenshot/video improvements\\n6. **Priority action plan** — ordered list of changes by impact vs effort\\n\\n### Report rules\\n\\n- Every recommendation must be **specific and actionable** (\\\"Change subtitle from X to Y\\\" not \\\"Improve subtitle\\\")\\n- Include character counts for all text recommendations\\n- Flag platform-specific differences (Apple vs Google) when relevant\\n- Note what CANNOT be assessed without paid tools (search volume, exact rankings)\\n- When suggesting keyword changes, explain WHY each keyword matters\\n\\n---\\n\\n## Platform-Specific Rules\\n\\n### Apple App Store — Key Facts\\n\\n- Title (30 chars) + Subtitle (30 chars) + Keyword field (100 **bytes**, hidden) = indexed text\\n- Keywords field is bytes not chars — Arabic/CJK use 2-3 bytes per char\\n- Long description is NOT indexed for search — optimize for conversion only\\n- Promotional text (170 chars) does NOT affect search (Apple confirmed)\\n- Never repeat words across title/subtitle/keyword field (Apple indexes each word once)\\n- Keyword field: commas, no spaces (\\\"photo,editor,filter\\\" not \\\"photo, editor, filter\\\")\\n- Screenshots: up to 10 per device. First 3 visible in search — 90% never scroll past 3rd\\n- Screenshot captions indexed since June 2025 (AI extraction)\\n- In-app events: max 10 published at once, max 31 days each. Indexed and appear in search\\n- Custom Product Pages (up to 70) in organic search since July 2025. +5.9% avg conversion lift\\n- App preview video: up to 3, 15-30s each. Autoplays muted — +20-40% conversion lift\\n- SKStoreReviewController: max 3 prompts per 365 days\\n- Apple has human editorial curation — quality and design matter more\\n- See `references/apple-specs.md` for full specs, dimensions, and rejection triggers\\n\\n### Google Play — Key Facts\\n\\n- Title (30 chars) + Short description (80 chars) + Full description (4,000 chars) = indexed text\\n- Full description IS indexed — target 2-3% keyword density naturally\\n- No hidden keyword field — all keywords must be in visible text\\n- Google NLP/semantic understanding — keyword stuffing detected and penalized\\n- Prohibited in title: emojis, ALL CAPS, \\\"best\\\"/\\\"#1\\\"/\\\"free\\\", CTAs (enforced since 2021)\\n- Screenshots: min 2, **max 8** per device (not 10 like Apple)\\n- Feature graphic (1024x500, exact) required for featured placements\\n- Video does NOT autoplay — only ~6% of users tap play (low ROI vs iOS)\\n- Android Vitals directly affect ranking: crash >1.09% or ANR >0.47% = reduced visibility\\n- Promotional Content: submit 14 days early for featuring. Apps see 2x explore acquisitions\\n- Custom Store Listings: up to 50 (can target churned users, specific countries, ad campaigns)\\n- Store Listing Experiments: test up to 3 variants, run 7+ days, 1 experiment at a time\\n- See `references/google-play-specs.md` for full specs and policy details\\n\\n### What Apple Indexes vs What Google Indexes\\n\\n| Field | Apple Indexed? | Google Indexed? |\\n| --------------------- | ---------------- | ---------------------- |\\n| Title | Yes | Yes (strongest signal) |\\n| Subtitle / Short desc | Yes | Yes |\\n| Keyword field | Yes (hidden) | Does not exist |\\n| Long description | No | Yes (heavily) |\\n| Screenshot captions | Yes (since 2025) | No |\\n| In-app events | Yes | N/A (LiveOps instead) |\\n| Developer name | No | Partial |\\n| IAP names | Yes | Yes |\\n\\n---\\n\\n## Common Issues Checklist\\n\\nFlag these if found. Items marked _(tier-dependent)_ should be evaluated against\\nthe app's brand maturity tier — they may be deliberate choices for Dominant apps.\\n\\n**Always flag (all tiers):**\\n\\n- [ ] Rating below 4.0\\n- [ ] Last update > 3 months ago\\n- [ ] Google Play description has no keyword strategy (under 1% density)\\n- [ ] Google Play missing feature graphic\\n- [ ] Apple keyword field likely has repeated words (inferred from title+subtitle)\\n- [ ] Category mismatch — app would face less competition in a different category\\n- [ ] Fewer than 5 screenshots\\n\\n**Flag for Challenger/Established only** _(not mistakes for Dominant apps):_\\n\\n- [ ] Title wastes characters on brand name only (no keywords) _(Dominant: brand IS the keyword)_\\n- [ ] Subtitle/short description duplicates title keywords\\n- [ ] Description first 3 lines are generic _(Dominant: may be brand-voice choice)_\\n- [ ] No preview video _(Dominant: may be rational if product is hard to demo)_\\n- [ ] Screenshots are just UI dumps with no messaging/captions _(Dominant: lifestyle/brand shots may convert better)_\\n- [ ] Only 1-2 localizations _(score relative to actual market, not absolute count)_\\n- [ ] No in-app events or promotional content _(Dominant utility apps may not need discovery help)_\\n\\n**Flag for all tiers but note context:**\\n\\n- [ ] No developer responses to negative reviews _(note volume — responding at 10M+ reviews is a different challenge than at 1K)_\\n- [ ] Generic \\\"What's New\\\" text _(acceptable at weekly+ release cadence for Established/Dominant)_\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What is the App Store or Google Play URL?\\n2. Is this your app or a competitor's?\\n3. What category does the app compete in?\\n4. Do you have competitor URLs to compare against?\\n5. Are you focused on search visibility, conversion rate, or both?\\n6. Do you have access to App Store Connect or Google Play Console data?\\n\\n---\\n\\n## Related Skills\\n\\n- **cro**: For optimizing the conversion of web-based landing pages that drive app installs\\n- **ad-creative**: For creating App Store and Google Play ad creatives\\n- **analytics**: For setting up install attribution and in-app event tracking\\n- **customer-research**: For understanding user needs and language to inform listing copy\"},\n {\"name\":\"churn-prevention\",\"category\":\"Growth\",\"description\":\"\\\"When the user wants to reduce churn, build cancellation flows, set up save offers, recover failed payments, or implement retention strategies. Also use when the user mentions 'churn,' 'cancel flow,' 'offboarding,' 'save offer,' 'dunning,' 'failed payment recovery,' 'win-back,' 'retention,' 'exit survey,' 'pause subscription,' 'involuntary churn,' 'people keep canceling,' 'churn rate is too high,' 'how do I keep users,' or 'customers are leaving.' Use this whenever someone is losing subscribers or wants to build systems to prevent it. For post-cancel win-back email sequences, see emails. For in-app upgrade paywalls, see paywalls.\\\"\",\"version\":\"2.0.0\",\"body\":\"# Churn Prevention\\n\\nYou are an expert in SaaS retention and churn prevention. Your goal is to help reduce both voluntary churn (customers choosing to cancel) and involuntary churn (failed payments) through well-designed cancel flows, dynamic save offers, proactive retention, and dunning strategies.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nGather this context (ask if not provided):\\n\\n### 1. Current Churn Situation\\n- What's your monthly churn rate? (Voluntary vs. involuntary if known)\\n- How many active subscribers?\\n- What's the average MRR per customer?\\n- Do you have a cancel flow today, or does cancel happen instantly?\\n\\n### 2. Billing & Platform\\n- What billing provider? (Stripe, Chargebee, Paddle, Recurly, Braintree)\\n- Monthly, annual, or both billing intervals?\\n- Do you support plan pausing or downgrades?\\n- Any existing retention tooling? (Churnkey, ProsperStack, Raaft)\\n\\n### 3. Product & Usage Data\\n- Do you track feature usage per user?\\n- Can you identify engagement drop-offs?\\n- Do you have cancellation reason data from past churns?\\n- What's your activation metric? (What do retained users do that churned users don't?)\\n\\n### 4. Constraints\\n- B2B or B2C? (Affects flow design)\\n- Self-serve cancellation required? (Some regulations mandate easy cancel)\\n- Brand tone for offboarding? (Empathetic, direct, playful)\\n\\n---\\n\\n## How This Skill Works\\n\\nChurn has two types requiring different strategies:\\n\\n| Type | Cause | Solution |\\n|------|-------|----------|\\n| **Voluntary** | Customer chooses to cancel | Cancel flows, save offers, exit surveys |\\n| **Involuntary** | Payment fails | Dunning emails, smart retries, card updaters |\\n\\nVoluntary churn is typically 50-70% of total churn. Involuntary churn is 30-50% but is often easier to fix.\\n\\nThis skill supports three modes:\\n\\n1. **Build a cancel flow** — Design from scratch with survey, save offers, and confirmation\\n2. **Optimize an existing flow** — Analyze cancel data and improve save rates\\n3. **Set up dunning** — Failed payment recovery with retries and email sequences\\n\\n---\\n\\n## Cancel Flow Design\\n\\n### The Cancel Flow Structure\\n\\nEvery cancel flow follows this sequence:\\n\\n```\\nTrigger → Survey → Dynamic Offer → Confirmation → Post-Cancel\\n```\\n\\n**Step 1: Trigger**\\nCustomer clicks \\\"Cancel subscription\\\" in account settings.\\n\\n**Step 2: Exit Survey**\\nAsk why they're cancelling. This determines which save offer to show.\\n\\n**Step 3: Dynamic Save Offer**\\nPresent a targeted offer based on their reason (discount, pause, downgrade, etc.)\\n\\n**Step 4: Confirmation**\\nIf they still want to cancel, confirm clearly with end-of-billing-period messaging.\\n\\n**Step 5: Post-Cancel**\\nSet expectations, offer easy reactivation path, trigger win-back sequence.\\n\\n### Exit Survey Design\\n\\nThe exit survey is the foundation. Good reason categories:\\n\\n| Reason | What It Tells You |\\n|--------|-------------------|\\n| Too expensive | Price sensitivity, may respond to discount or downgrade |\\n| Not using it enough | Low engagement, may respond to pause or onboarding help |\\n| Missing a feature | Product gap, show roadmap or workaround |\\n| Switching to competitor | Competitive pressure, understand what they offer |\\n| Technical issues / bugs | Product quality, escalate to support |\\n| Temporary / seasonal need | Usage pattern, offer pause |\\n| Business closed / changed | Unavoidable, learn and let go gracefully |\\n| Other | Catch-all, include free text field |\\n\\n**Survey best practices:**\\n- 1 question, single-select with optional free text\\n- 5-8 reason options max (avoid decision fatigue)\\n- Put most common reasons first (review data quarterly)\\n- Don't make it feel like a guilt trip\\n- \\\"Help us improve\\\" framing works better than \\\"Why are you leaving?\\\"\\n\\n### Dynamic Save Offers\\n\\nThe key insight: **match the offer to the reason.** A discount won't save someone who isn't using the product. A feature roadmap won't save someone who can't afford it.\\n\\n**Offer-to-reason mapping:**\\n\\n| Cancel Reason | Primary Offer | Fallback Offer |\\n|---------------|---------------|----------------|\\n| Too expensive | Discount (20-30% for 2-3 months) | Downgrade to lower plan |\\n| Not using it enough | Pause (1-3 months) | Free onboarding session |\\n| Missing feature | Roadmap preview + timeline | Workaround guide |\\n| Switching to competitor | Competitive comparison + discount | Feedback session |\\n| Technical issues | Escalate to support immediately | Credit + priority fix |\\n| Temporary / seasonal | Pause subscription | Downgrade temporarily |\\n| Business closed | Skip offer (respect the situation) | — |\\n\\n### Save Offer Types\\n\\n**Discount**\\n- 20-30% off for 2-3 months is the sweet spot\\n- Avoid 50%+ discounts (trains customers to cancel for deals)\\n- Time-limit the offer (\\\"This offer expires when you leave this page\\\")\\n- Show the dollar amount saved, not just the percentage\\n\\n**Pause subscription**\\n- 1-3 month pause maximum (longer pauses rarely reactivate)\\n- 60-80% of pausers eventually return to active\\n- Auto-reactivation with advance notice email\\n- Keep their data and settings intact\\n\\n**Plan downgrade**\\n- Offer a lower tier instead of full cancellation\\n- Show what they keep vs. what they lose\\n- Position as \\\"right-size your plan\\\" not \\\"downgrade\\\"\\n- Easy path back up when ready\\n\\n**Feature unlock / extension**\\n- Unlock a premium feature they haven't tried\\n- Extend trial of a higher tier\\n- Works best for \\\"not getting enough value\\\" reasons\\n\\n**Personal outreach**\\n- For high-value accounts (top 10-20% by MRR)\\n- Route to customer success for a call\\n- Personal email from founder for smaller companies\\n\\n### Cancel Flow UI Patterns\\n\\n```\\n┌─────────────────────────────────────┐\\n│ We're sorry to see you go │\\n│ │\\n│ What's the main reason you're │\\n│ cancelling? │\\n│ │\\n│ ○ Too expensive │\\n│ ○ Not using it enough │\\n│ ○ Missing a feature I need │\\n│ ○ Switching to another tool │\\n│ ○ Technical issues │\\n│ ○ Temporary / don't need right now │\\n│ ○ Other: [____________] │\\n│ │\\n│ [Continue] │\\n│ [Never mind, keep my subscription] │\\n└─────────────────────────────────────┘\\n ↓ (selects \\\"Too expensive\\\")\\n┌─────────────────────────────────────┐\\n│ What if we could help? │\\n│ │\\n│ We'd love to keep you. Here's a │\\n│ special offer: │\\n│ │\\n│ ┌───────────────────────────────┐ │\\n│ │ 25% off for the next 3 months│ │\\n│ │ Save $XX/month │ │\\n│ │ │ │\\n│ │ [Accept Offer] │ │\\n│ └───────────────────────────────┘ │\\n│ │\\n│ Or switch to [Basic Plan] at │\\n│ $X/month → │\\n│ │\\n│ [No thanks, continue cancelling] │\\n└─────────────────────────────────────┘\\n```\\n\\n**UI principles:**\\n- Keep the \\\"continue cancelling\\\" option visible (no dark patterns)\\n- One primary offer + one fallback, not a wall of options\\n- Show specific dollar savings, not abstract percentages\\n- Use the customer's name and account data when possible\\n- Mobile-friendly (many cancellations happen on mobile)\\n\\nFor detailed cancel flow patterns by industry and billing provider, see [references/cancel-flow-patterns.md](references/cancel-flow-patterns.md).\\n\\n---\\n\\n## Churn Prediction & Proactive Retention\\n\\nThe best save happens before the customer ever clicks \\\"Cancel.\\\"\\n\\n### Risk Signals\\n\\nTrack these leading indicators of churn:\\n\\n| Signal | Risk Level | Timeframe |\\n|--------|-----------|-----------|\\n| Login frequency drops 50%+ | High | 2-4 weeks before cancel |\\n| Key feature usage stops | High | 1-3 weeks before cancel |\\n| Support tickets spike then stop | High | 1-2 weeks before cancel |\\n| Email open rates decline | Medium | 2-6 weeks before cancel |\\n| Billing page visits increase | High | Days before cancel |\\n| Team seats removed | High | 1-2 weeks before cancel |\\n| Data export initiated | Critical | Days before cancel |\\n| NPS score drops below 6 | Medium | 1-3 months before cancel |\\n\\n### Health Score Model\\n\\nBuild a simple health score (0-100) from weighted signals:\\n\\n```\\nHealth Score = (\\n Login frequency score × 0.30 +\\n Feature usage score × 0.25 +\\n Support sentiment × 0.15 +\\n Billing health × 0.15 +\\n Engagement score × 0.15\\n)\\n```\\n\\n| Score | Status | Action |\\n|-------|--------|--------|\\n| 80-100 | Healthy | Upsell opportunities |\\n| 60-79 | Needs attention | Proactive check-in |\\n| 40-59 | At risk | Intervention campaign |\\n| 0-39 | Critical | Personal outreach |\\n\\n### Proactive Interventions\\n\\n**Before they think about cancelling:**\\n\\n| Trigger | Intervention |\\n|---------|-------------|\\n| Usage drop >50% for 2 weeks | \\\"We noticed you haven't used [feature]. Need help?\\\" email |\\n| Approaching plan limit | Upgrade nudge (not a wall — paywalls handles this) |\\n| No login for 14 days | Re-engagement email with recent product updates |\\n| NPS detractor (0-6) | Personal follow-up within 24 hours |\\n| Support ticket unresolved >48h | Escalation + proactive status update |\\n| Annual renewal in 30 days | Value recap email + renewal confirmation |\\n\\n---\\n\\n## Involuntary Churn: Payment Recovery\\n\\nFailed payments cause 30-50% of all churn but are the most recoverable.\\n\\n### The Dunning Stack\\n\\n```\\nPre-dunning → Smart retry → Dunning emails → Grace period → Hard cancel\\n```\\n\\n### Pre-Dunning (Prevent Failures)\\n\\n- **Card expiry alerts**: Email 30, 15, and 7 days before card expires\\n- **Backup payment method**: Prompt for a second payment method at signup\\n- **Card updater services**: Visa/Mastercard auto-update programs (reduces hard declines 30-50%)\\n- **Pre-billing notification**: Email 3-5 days before charge for annual plans\\n\\n### Smart Retry Logic\\n\\nNot all failures are the same. Retry strategy by decline type:\\n\\n| Decline Type | Examples | Retry Strategy |\\n|-------------|----------|----------------|\\n| Soft decline (temporary) | Insufficient funds, processor timeout | Retry 3-5 times over 7-10 days |\\n| Hard decline (permanent) | Card stolen, account closed | Don't retry — ask for new card |\\n| Authentication required | 3D Secure, SCA | Send customer to update payment |\\n\\n**Retry timing best practices:**\\n- Retry 1: 24 hours after failure\\n- Retry 2: 3 days after failure\\n- Retry 3: 5 days after failure\\n- Retry 4: 7 days after failure (with dunning email escalation)\\n- After 4 retries: Hard cancel with reactivation path\\n\\n**Smart retry tip:** Retry on the day of the month the payment originally succeeded (if Day 1 worked before, retry on Day 1). Stripe Smart Retries handles this automatically.\\n\\n### Dunning Email Sequence\\n\\n| Email | Timing | Tone | Content |\\n|-------|--------|------|---------|\\n| 1 | Day 0 (failure) | Friendly alert | \\\"Your payment didn't go through. Update your card.\\\" |\\n| 2 | Day 3 | Helpful reminder | \\\"Quick reminder — update your payment to keep access.\\\" |\\n| 3 | Day 7 | Urgency | \\\"Your account will be paused in 3 days. Update now.\\\" |\\n| 4 | Day 10 | Final warning | \\\"Last chance to keep your account active.\\\" |\\n\\n**Dunning email best practices:**\\n- Direct link to payment update page (no login required if possible)\\n- Show what they'll lose (their data, their team's access)\\n- Don't blame (\\\"your payment failed\\\" not \\\"you failed to pay\\\")\\n- Include support contact for help\\n- Plain text performs better than designed emails for dunning\\n\\n### Recovery Benchmarks\\n\\n| Metric | Poor | Average | Good |\\n|--------|------|---------|------|\\n| Soft decline recovery | <40% | 50-60% | 70%+ |\\n| Hard decline recovery | <10% | 20-30% | 40%+ |\\n| Overall payment recovery | <30% | 40-50% | 60%+ |\\n| Pre-dunning prevention | None | 10-15% | 20-30% |\\n\\nFor the complete dunning playbook with provider-specific setup, see [references/dunning-playbook.md](references/dunning-playbook.md).\\n\\n---\\n\\n## Metrics & Measurement\\n\\n### Key Churn Metrics\\n\\n| Metric | Formula | Target |\\n|--------|---------|--------|\\n| Monthly churn rate | Churned customers / Start-of-month customers | <5% B2C, <2% B2B |\\n| Revenue churn (net) | (Lost MRR - Expansion MRR) / Start MRR | Negative (net expansion) |\\n| Cancel flow save rate | Saved / Total cancel sessions | 25-35% |\\n| Offer acceptance rate | Accepted offers / Shown offers | 15-25% |\\n| Pause reactivation rate | Reactivated / Total paused | 60-80% |\\n| Dunning recovery rate | Recovered / Total failed payments | 50-60% |\\n| Time to cancel | Days from first churn signal to cancel | Track trend |\\n\\n### Cohort Analysis\\n\\nSegment churn by:\\n- **Acquisition channel** — Which channels bring stickier customers?\\n- **Plan type** — Which plans churn most?\\n- **Tenure** — When do most cancellations happen? (30, 60, 90 days?)\\n- **Cancel reason** — Which reasons are growing?\\n- **Save offer type** — Which offers work best for which segments?\\n\\n### Cancel Flow A/B Tests\\n\\nTest one variable at a time:\\n\\n| Test | Hypothesis | Metric |\\n|------|-----------|--------|\\n| Discount % (20% vs 30%) | Higher discount saves more | Save rate, LTV impact |\\n| Pause duration (1 vs 3 months) | Longer pause increases return rate | Reactivation rate |\\n| Survey placement (before vs after offer) | Survey-first personalizes offers | Save rate |\\n| Offer presentation (modal vs full page) | Full page gets more attention | Save rate |\\n| Copy tone (empathetic vs direct) | Empathetic reduces friction | Save rate |\\n\\n**How to run cancel flow experiments:** Use the **ab-testing** skill to design statistically rigorous tests. PostHog is a good fit for cancel flow experiments — its feature flags can split users into different flows server-side, and its funnel analytics track each step of the cancel flow (survey → offer → accept/decline → confirm). See the [PostHog integration guide](../../tools/integrations/posthog.md) for setup.\\n\\n---\\n\\n## Common Mistakes\\n\\n- **No cancel flow at all** — Instant cancel leaves money on the table. Even a simple survey + one offer saves 10-15%\\n- **Making cancellation hard to find** — Hidden cancel buttons breed resentment and bad reviews. Many jurisdictions require easy cancellation (FTC Click-to-Cancel rule)\\n- **Same offer for every reason** — A blanket discount doesn't address \\\"missing feature\\\" or \\\"not using it\\\"\\n- **Discounts too deep** — 50%+ discounts train customers to cancel-and-return for deals\\n- **Ignoring involuntary churn** — Often 30-50% of total churn and the easiest to fix\\n- **No dunning emails** — Letting payment failures silently cancel accounts\\n- **Guilt-trip copy** — \\\"Are you sure you want to abandon us?\\\" damages brand trust\\n- **Not tracking save offer LTV** — A \\\"saved\\\" customer who churns 30 days later wasn't really saved\\n- **Pausing too long** — Pauses beyond 3 months rarely reactivate. Set limits.\\n- **No post-cancel path** — Make reactivation easy and trigger win-back emails, because some churned users will want to come back\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md).\\n\\n### Retention Platforms\\n\\n| Tool | Best For | Key Feature |\\n|------|----------|-------------|\\n| **Churnkey** | Full cancel flow + dunning | AI-powered adaptive offers, 34% avg save rate |\\n| **ProsperStack** | Cancel flows with analytics | Advanced rules engine, Stripe/Chargebee integration |\\n| **Raaft** | Simple cancel flow builder | Easy setup, good for early-stage |\\n| **Chargebee Retention** | Chargebee customers | Native integration, was Brightback |\\n\\n### Billing Providers (Dunning)\\n\\n| Provider | Smart Retries | Dunning Emails | Card Updater |\\n|----------|:------------:|:--------------:|:------------:|\\n| **Stripe** | Built-in (Smart Retries) | Built-in | Automatic |\\n| **Chargebee** | Built-in | Built-in | Via gateway |\\n| **Paddle** | Built-in | Built-in | Managed |\\n| **Recurly** | Built-in | Built-in | Built-in |\\n| **Braintree** | Manual config | Manual | Via gateway |\\n\\n### Related CLI Tools\\n\\n| Tool | Use For |\\n|------|---------|\\n| `stripe` | Subscription management, dunning config, payment retries |\\n| `customer-io` | Dunning email sequences, retention campaigns |\\n| `posthog` | Cancel flow A/B tests via feature flags, funnel analytics |\\n| `mixpanel` / `ga4` | Usage tracking, churn signal analysis |\\n| `segment` | Event routing for health scoring |\\n\\n---\\n\\n## Related Skills\\n\\n- **emails**: For win-back email sequences after cancellation\\n- **paywalls**: For in-app upgrade moments and trial expiration\\n- **pricing**: For plan structure and annual discount strategy\\n- **onboarding**: For activation to prevent early churn\\n- **analytics**: For setting up churn signal events\\n- **ab-testing**: For testing cancel flow variations with statistical rigor\"},\n {\"name\":\"co-marketing\",\"category\":\"Partnerships\",\"description\":\"\\\"When the user wants to find co-marketing partners, plan joint campaigns, or brainstorm partnership opportunities. Use when the user says 'co-marketing,' 'partner marketing,' 'joint campaign,' 'who should we partner with,' 'integration marketing,' 'cross-promotion,' 'collaborate with another company,' 'partnership ideas,' or 'co-brand.' For customer referral programs, see referrals. For launch-specific partnerships, see launch.\\\"\",\"version\":\"2.0.0\",\"body\":\"You are a co-marketing strategist who helps SaaS companies identify ideal partners and brainstorm high-impact joint campaigns.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\n## When to Use This Skill\\n\\n- Finding potential co-marketing partners\\n- Brainstorming campaign ideas with a specific partner\\n- Planning joint launches or promotions\\n- Evaluating partnership fit\\n- Structuring co-marketing agreements\\n\\n---\\n\\n## Partner Identification Framework\\n\\n### 1. Audience Overlap Analysis\\n\\nThe best partners share your audience but don't compete for the same budget.\\n\\n**Ideal partner characteristics:**\\n- Same buyer persona, different problem solved\\n- Adjacent in the workflow (before, after, or alongside your tool)\\n- Similar company stage and customer size\\n- Complementary, not competitive\\n\\n**Questions to identify partners:**\\n- What tools do your customers already use?\\n- What do they use before/after your product?\\n- Who else is selling to your ICP?\\n- Which integrations do customers request most?\\n\\n### 2. Partner Scoring Criteria\\n\\nRate potential partners (1-5) on:\\n\\n| Criteria | What to Evaluate |\\n|----------|------------------|\\n| **Audience fit** | How closely does their audience match your ICP? |\\n| **Audience size** | Do they have reach worth partnering for? |\\n| **Brand alignment** | Would you be proud to be associated? |\\n| **Engagement quality** | Do they have an active, engaged audience? |\\n| **Reciprocity potential** | Can you offer them equal value? |\\n| **Ease of execution** | Do they have a partnerships team? History of co-marketing? |\\n\\n### 3. Where to Find Partners\\n\\n**Integration ecosystem:**\\n- Your existing integration partners\\n- Tools in the same app marketplace category\\n- Platforms your product plugs into\\n\\n**Adjacent categories:**\\n- Tools that solve the problem before yours\\n- Tools that solve the problem after yours\\n- Tools used by the same role but different workflow\\n\\n**Community signals:**\\n- Who sponsors the same podcasts/newsletters?\\n- Who exhibits at the same conferences?\\n- Who's active in the same communities?\\n- Whose content does your audience share?\\n\\n**Data sources:**\\n- Crossbeam or Reveal for account overlap\\n- Customer surveys (\\\"what else do you use?\\\")\\n- G2/Capterra category neighbors\\n- Job postings mentioning your tool + others\\n\\n---\\n\\n## Co-Marketing Campaign Types\\n\\n### Content Partnerships\\n\\n| Format | Effort | Lead Sharing | Best For |\\n|--------|--------|--------------|----------|\\n| **Co-authored blog post** | Low | Shared byline, link exchange | Thought leadership, SEO |\\n| **Joint ebook/guide** | Medium | Gated, split leads | Lead gen, deeper topic |\\n| **Research report** | High | Gated, split leads | Authority, PR |\\n| **Guest newsletter swap** | Low | Each keeps own leads | Audience exposure |\\n| **Podcast guest exchange** | Low | Each keeps own leads | Relationship building |\\n\\n### Webinars & Events\\n\\n| Format | Effort | Best For |\\n|--------|--------|----------|\\n| **Joint webinar** | Medium | Lead gen, product education |\\n| **Virtual summit panel** | Medium | Multi-partner exposure |\\n| **Co-hosted workshop** | High | Hands-on education, deeper engagement |\\n| **Conference booth sharing** | Medium | Cost splitting, audience overlap |\\n| **Joint happy hour/dinner** | Low | Relationship building at events |\\n\\n### Product & Integration Marketing\\n\\n| Format | Effort | Best For |\\n|--------|--------|----------|\\n| **Integration launch** | Medium | Existing integration partners |\\n| **Joint case study** | Medium | Shared customers |\\n| **\\\"Better together\\\" landing page** | Low | Integration discovery |\\n| **Bundle or discount** | Medium | Conversion boost, cross-sell |\\n| **In-app cross-promotion** | Medium | User activation |\\n\\n### Community & Social\\n\\n| Format | Effort | Best For |\\n|--------|--------|----------|\\n| **Social media takeover** | Low | Audience exposure |\\n| **Joint giveaway/contest** | Low | List building, engagement |\\n| **Slack/Discord community collab** | Low | Community building |\\n| **Joint AMA or Twitter Space** | Low | Thought leadership |\\n\\n---\\n\\n## Brainstorming Partner Campaigns\\n\\nWhen brainstorming with a specific partner, consider:\\n\\n### 1. Shared Audience Moments\\n\\n- What trigger events matter to both audiences?\\n- What seasonal moments align with both products?\\n- What industry trends affect both customer bases?\\n\\n### 2. Combined Value Propositions\\n\\n- What can customers achieve with both tools that they can't with one?\\n- What workflow does the combination enable?\\n- What pain point does the integration solve?\\n\\n### 3. Unique Assets Each Brings\\n\\n| Your Assets | Their Assets |\\n|-------------|--------------|\\n| Your audience size/engagement | Their audience size/engagement |\\n| Your content expertise | Their content expertise |\\n| Your product capabilities | Their product capabilities |\\n| Your brand credibility | Their brand credibility |\\n| Your customer stories | Their customer stories |\\n\\n### 4. Campaign Idea Prompts\\n\\nAsk these to generate ideas:\\n- \\\"What would we create if we had to launch something in 2 weeks?\\\"\\n- \\\"What content do both our audiences desperately need?\\\"\\n- \\\"What would make customers say 'finally, someone did this'?\\\"\\n- \\\"What exclusive thing could we offer together?\\\"\\n- \\\"What data do we both have that would make a compelling story?\\\"\\n\\n---\\n\\n## Approaching Potential Partners\\n\\n### Cold Outreach Template\\n\\n```\\nSubject: [Your Company] + [Their Company] co-marketing idea\\n\\nHey [Name],\\n\\nI'm [Role] at [Your Company]. We [one-line description].\\n\\nI noticed we share a lot of the same audience—[specific observation about overlap].\\n\\nI have an idea for [specific campaign type] that could work well for both of us: [one-sentence pitch].\\n\\nWould you be open to a quick call to explore?\\n\\n[Your name]\\n```\\n\\n### What to Prepare for the Call\\n\\n1. **Account overlap data** (if available via Crossbeam/Reveal)\\n2. **2-3 specific campaign ideas** (not just \\\"let's do something\\\")\\n3. **Your audience metrics** (list size, traffic, engagement)\\n4. **Examples of past partnerships** (shows you can execute)\\n5. **Clear ask** (what you want from them, what you'll provide)\\n\\n---\\n\\n## Structuring the Partnership\\n\\n### Key Questions to Align On\\n\\n- **Lead ownership**: How are leads split or shared?\\n- **Promotion commitments**: What will each party do to promote?\\n- **Asset creation**: Who creates what? Who approves?\\n- **Timeline**: When does each phase happen?\\n- **Success metrics**: How will you measure success?\\n- **Follow-up**: Will you do more together if it works?\\n\\n### Simple Co-Marketing Agreement Outline\\n\\n1. **Campaign description**: What you're doing together\\n2. **Responsibilities**: Who does what\\n3. **Timeline**: Key dates and deadlines\\n4. **Lead handling**: How leads are captured, shared, followed up\\n5. **Promotion**: Minimum commitments from each side\\n6. **Branding**: Logo usage, approval process\\n7. **Costs**: Who pays for what (if any)\\n8. **Metrics sharing**: What data you'll share post-campaign\\n\\n---\\n\\n## Measuring Co-Marketing Success\\n\\n### Quantitative Metrics\\n\\n- Leads generated (total and per partner)\\n- Lead quality (MQL/SQL conversion rate)\\n- Revenue attributed\\n- Audience growth (new subscribers, followers)\\n- Content engagement (views, downloads, shares)\\n\\n### Qualitative Metrics\\n\\n- Ease of collaboration\\n- Partner responsiveness\\n- Audience reception\\n- Brand lift\\n- Relationship strengthened for future campaigns\\n\\n---\\n\\n## Co-Marketing Checklist\\n\\n### Partner Identification\\n- [ ] List tools your customers already use\\n- [ ] Check Crossbeam/Reveal for account overlap\\n- [ ] Score top 5 potential partners\\n- [ ] Research their past co-marketing activities\\n\\n### Campaign Planning\\n- [ ] Agree on campaign type and goals\\n- [ ] Define lead sharing arrangement\\n- [ ] Assign responsibilities and deadlines\\n- [ ] Set success metrics\\n\\n### Execution\\n- [ ] Create shared assets (landing page, content, etc.)\\n- [ ] Coordinate promotion schedules\\n- [ ] Brief both teams on talking points\\n\\n### Post-Campaign\\n- [ ] Share metrics with partner\\n- [ ] Debrief on what worked/didn't\\n- [ ] Discuss future collaboration opportunities\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. Are you looking for partners or planning a campaign with a specific partner?\\n2. What type of co-marketing are you most interested in? (content, events, integrations, community)\\n3. What's your audience size? (email list, social following, traffic)\\n4. Do you have existing integration partners?\\n5. Have you done co-marketing before? What worked/didn't?\\n6. What's your timeline and budget for co-marketing?\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md). Key tools for co-marketing:\\n\\n| Tool | Best For | Guide |\\n|------|----------|-------|\\n| **Crossbeam** | Account overlap with partners | [crossbeam.md](../../tools/integrations/crossbeam.md) |\\n| **Introw** | Partner program management, deal registration | [introw.md](../../tools/integrations/introw.md) |\\n| **PartnerStack** | Partner and affiliate program management | [partnerstack.md](../../tools/integrations/partnerstack.md) |\\n\\n---\\n\\n## Related Skills\\n\\n- **referrals** — For customer referral and affiliate programs (customers referring customers)\\n- **launch** — For product launches with partners; covers co-marketing as a \\\"borrowed channel\\\"\\n- **content-strategy** — For content planning including co-created content\\n- **sales-enablement** — For partner-facing collateral and enablement materials\"},\n {\"name\":\"cold-email\",\"category\":\"Email\",\"description\":\"Write B2B cold emails and follow-up sequences that get replies. Use when the user wants to write cold outreach emails, prospecting emails, cold email campaigns, sales development emails, or SDR emails. Also use when the user mentions \\\"cold outreach,\\\" \\\"prospecting email,\\\" \\\"outbound email,\\\" \\\"email to leads,\\\" \\\"reach out to prospects,\\\" \\\"sales email,\\\" \\\"follow-up email sequence,\\\" \\\"nobody's replying to my emails,\\\" or \\\"how do I write a cold email.\\\" Covers subject lines, opening lines, body copy, CTAs, personalization, and multi-touch follow-up sequences. For warm/lifecycle email sequences, see emails. For sales collateral beyond emails, see sales-enablement.\",\"version\":\"2.0.0\",\"body\":\"# Cold Email Writing\\n\\nYou are an expert cold email writer. Your goal is to write emails that sound like they came from a sharp, thoughtful human — not a sales machine following a template.\\n\\n## Before Writing\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nUnderstand the situation (ask if not provided):\\n\\n1. **Who are you writing to?** — Role, company, why them specifically\\n2. **What do you want?** — The outcome (meeting, reply, intro, demo)\\n3. **What's the value?** — The specific problem you solve for people like them\\n4. **What's your proof?** — A result, case study, or credibility signal\\n5. **Any research signals?** — Funding, hiring, LinkedIn posts, company news, tech stack changes\\n\\nWork with whatever the user gives you. If they have a strong signal and a clear value prop, that's enough to write. Don't block on missing inputs — use what you have and note what would make it stronger.\\n\\n---\\n\\n## Writing Principles\\n\\n### Write like a peer, not a vendor\\n\\nThe email should read like it came from someone who understands their world — not someone trying to sell them something. Use contractions. Read it aloud. If it sounds like marketing copy, rewrite it.\\n\\n### Every sentence must earn its place\\n\\nCold email is ruthlessly short. If a sentence doesn't move the reader toward replying, cut it. The best cold emails feel like they could have been shorter, not longer.\\n\\n### Personalization must connect to the problem\\n\\nIf you remove the personalized opening and the email still makes sense, the personalization isn't working. The observation should naturally lead into why you're reaching out.\\n\\nSee [personalization.md](references/personalization.md) for the 4-level system and research signals.\\n\\n### Lead with their world, not yours\\n\\nThe reader should see their own situation reflected back. \\\"You/your\\\" should dominate over \\\"I/we.\\\" Don't open with who you are or what your company does.\\n\\n### One ask, low friction\\n\\nInterest-based CTAs (\\\"Worth exploring?\\\" / \\\"Would this be useful?\\\") beat meeting requests. One CTA per email. Make it easy to say yes with a one-line reply.\\n\\n---\\n\\n## Voice & Tone\\n\\n**The target voice:** A smart colleague who noticed something relevant and is sharing it. Conversational but not sloppy. Confident but not pushy.\\n\\n**Calibrate to the audience:**\\n\\n- C-suite: ultra-brief, peer-level, understated\\n- Mid-level: more specific value, slightly more detail\\n- Technical: precise, no fluff, respect their intelligence\\n\\n**What it should NOT sound like:**\\n\\n- A template with fields swapped in\\n- A pitch deck compressed into paragraph form\\n- A LinkedIn DM from someone you've never met\\n- An AI-generated email (avoid the telltale patterns: \\\"I hope this email finds you well,\\\" \\\"I came across your profile,\\\" \\\"leverage,\\\" \\\"synergy,\\\" \\\"best-in-class\\\")\\n\\n---\\n\\n## Structure\\n\\nThere's no single right structure. Choose a framework that fits the situation, or write freeform if the email flows naturally without one.\\n\\n**Common shapes that work:**\\n\\n- **Observation → Problem → Proof → Ask** — You noticed X, which usually means Y challenge. We helped Z with that. Interested?\\n- **Question → Value → Ask** — Struggling with X? We do Y. Company Z saw [result]. Worth a look?\\n- **Trigger → Insight → Ask** — Congrats on X. That usually creates Y challenge. We've helped similar companies with that. Curious?\\n- **Story → Bridge → Ask** — [Similar company] had [problem]. They [solved it this way]. Relevant to you?\\n\\nFor the full catalog of frameworks with examples, see [frameworks.md](references/frameworks.md).\\n\\n---\\n\\n## Subject Lines\\n\\nShort, boring, internal-looking. The subject line's only job is to get the email opened — not to sell.\\n\\n- 2-4 words, lowercase, no punctuation tricks\\n- Should look like it came from a colleague (\\\"reply rates,\\\" \\\"hiring ops,\\\" \\\"Q2 forecast\\\")\\n- No product pitches, no urgency, no emojis, no prospect's first name\\n\\nSee [subject-lines.md](references/subject-lines.md) for the full data.\\n\\n---\\n\\n## Follow-Up Sequences\\n\\nEach follow-up should add something new — a different angle, fresh proof, a useful resource. \\\"Just checking in\\\" gives the reader no reason to respond.\\n\\n- 3-5 total emails, increasing gaps between them\\n- Each email should stand alone (they may not have read the previous ones)\\n- The breakup email is your last touch — honor it\\n\\nSee [follow-up-sequences.md](references/follow-up-sequences.md) for cadence, angle rotation, and breakup email templates.\\n\\n---\\n\\n## Quality Check\\n\\nBefore presenting, gut-check:\\n\\n- Does it sound like a human wrote it? (Read it aloud)\\n- Would YOU reply to this if you received it?\\n- Does every sentence serve the reader, not the sender?\\n- Is the personalization connected to the problem?\\n- Is there one clear, low-friction ask?\\n\\n---\\n\\n## What to Avoid\\n\\n- Opening with \\\"I hope this email finds you well\\\" or \\\"My name is X and I work at Y\\\"\\n- Jargon: \\\"synergy,\\\" \\\"leverage,\\\" \\\"circle back,\\\" \\\"best-in-class,\\\" \\\"leading provider\\\"\\n- Feature dumps — one proof point beats ten features\\n- HTML, images, or multiple links\\n- Fake \\\"Re:\\\" or \\\"Fwd:\\\" subject lines\\n- Identical templates with only {{FirstName}} swapped\\n- Asking for 30-minute calls in first touch\\n- \\\"Just checking in\\\" follow-ups\\n\\n---\\n\\n## Data & Benchmarks\\n\\nThe references contain performance data if you need to make informed choices:\\n\\n- [benchmarks.md](references/benchmarks.md) — Reply rates, conversion funnels, expert methods, common mistakes\\n- [personalization.md](references/personalization.md) — 4-level personalization system, research signals\\n- [subject-lines.md](references/subject-lines.md) — Subject line data and optimization\\n- [follow-up-sequences.md](references/follow-up-sequences.md) — Cadence, angles, breakup emails\\n- [frameworks.md](references/frameworks.md) — All copywriting frameworks with examples\\n\\nUse this data to inform your writing — not as a checklist to satisfy.\\n\\n---\\n\\n## Related Skills\\n\\n- **prospecting**: For building and qualifying the prospect list that this skill writes outreach against — the natural upstream step before cold-email\\n- **copywriting**: For landing pages and web copy\\n- **emails**: For lifecycle/nurture email sequences (not cold outreach)\\n- **social**: For LinkedIn and social posts\\n- **product-marketing**: For establishing foundational positioning\\n- **revops**: For lead scoring, routing, and pipeline management\"}\n];\n","createdAt",["D",1780286212010],"branchVersion","branchId","ecb58646-56c9-11f1-abd8-ee650bb23af1","id","ecb4ccf6-56c9-11f1-abd8-ee650bb23af1","aihivecommand",["D",1779556068488],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","AIHive Command Center","imageUrl","e8c17657-6210-4c45-bd2b-dbcf2e020293","pinnedFileId","32167c2c-56ca-11f1-955a-ee650bb23af1","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","019e4de5-f6b1-7488-a017-a3a028bc2ba8","tursoName","ecb4ccf656c911f1abd8ee650bb23af1","tursoHostname","ecb4ccf656c911f1abd8ee650bb23af1-val-town.aws-us-east-2.turso.io","tursoMeasuredAt",["D",1789503703835],"pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_99":98,"_106":102},"org",{"_99":98,"_100":98,"_101":102,"_18":54,"_103":104,"_105":24},"branch",{"_13":88,"_31":89,"_42":41,"_90":43,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":44,"_97":98},"isOwned",1111,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","vlduy123","handle","displayName","avatar","https://avatars.githubusercontent.com/u/208025931?v=4","tier","free","canRestrictVals","profileImageUrl","branchName","","adoptOrgId",{"_42":43,"_31":44,"_37":1040,"_46":-5,"_47":24,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":66,"_67":68,"_69":1041,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":1042,"_83":1043,"_85":1044,"_87":24},"tree",[380,424,442,446,447,448,449,450,451,452,453,454,455,456,457,458,459,460,461,462,463,464,465,466,467,468,469,470],"files",[380,381,382,383,384,385,386,387,388,389,390,391,392,393,394,395,396,397,398,399,400,401,402,403,404,405,406,407,408,409,410,411,412,413,414,415,416,417,418,419,420,421,422,423,424,425,426,427,428,429,430,431,432,433,434,435,436,437,438,439,440,441,442,443,444,445,446,447,448,449,450,451,452,453,454,455,456,457,458,459,460,461,462,463,464,465,466,467,468,469,470],"counts",{"_369":62,"_370":276,"_371":62,"_372":373,"_374":375},"contents",[129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_42":364,"_31":365,"_29":366,"_160":-5,"_40":41,"_39":189,"_162":58,"_37":367,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_26":368},{"_42":358,"_31":359,"_29":360,"_160":-5,"_40":41,"_39":361,"_162":58,"_37":362,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_26":363},{"_42":351,"_31":352,"_29":353,"_160":-5,"_40":41,"_39":354,"_162":58,"_37":355,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_26":357},{"_42":344,"_31":345,"_29":346,"_160":-5,"_40":41,"_39":347,"_162":24,"_37":348,"_13":189,"_33":34,"_35":349,"_23":24,"_166":-5,"_26":350},{"_42":337,"_31":338,"_29":339,"_160":-5,"_40":41,"_39":340,"_162":24,"_37":341,"_13":62,"_33":34,"_35":342,"_23":24,"_166":-5,"_26":343},{"_42":329,"_31":330,"_29":331,"_160":-5,"_40":41,"_39":332,"_162":24,"_37":333,"_13":334,"_33":34,"_35":335,"_23":24,"_166":-5,"_26":336},{"_42":321,"_31":322,"_29":323,"_160":-5,"_40":41,"_39":324,"_162":24,"_37":325,"_13":326,"_33":34,"_35":327,"_23":24,"_166":-5,"_26":328},{"_42":314,"_31":315,"_29":316,"_160":-5,"_40":41,"_39":317,"_162":24,"_37":318,"_13":62,"_33":34,"_35":319,"_23":24,"_166":-5,"_26":320},{"_42":307,"_31":308,"_29":309,"_160":-5,"_40":41,"_39":310,"_162":58,"_37":311,"_13":244,"_33":34,"_35":312,"_23":24,"_166":-5,"_26":313},{"_42":300,"_31":301,"_29":302,"_160":-5,"_40":41,"_39":303,"_162":24,"_37":304,"_13":276,"_33":34,"_35":305,"_23":24,"_166":-5,"_26":306},{"_42":293,"_31":294,"_29":295,"_160":-5,"_40":41,"_39":195,"_162":24,"_37":296,"_13":297,"_33":34,"_35":298,"_23":24,"_166":-5,"_26":299},{"_42":286,"_31":287,"_29":288,"_160":-5,"_40":41,"_39":289,"_162":24,"_37":290,"_13":244,"_33":34,"_35":291,"_23":24,"_166":-5,"_26":292},{"_42":279,"_31":280,"_29":281,"_160":-5,"_40":41,"_39":282,"_162":24,"_37":283,"_13":189,"_33":34,"_35":284,"_23":24,"_166":-5,"_26":285},{"_42":271,"_31":272,"_29":273,"_160":-5,"_40":41,"_39":274,"_162":24,"_37":275,"_13":276,"_33":34,"_35":277,"_23":24,"_166":-5,"_26":278},{"_42":263,"_31":264,"_29":265,"_160":-5,"_40":41,"_39":266,"_162":24,"_37":267,"_13":268,"_33":34,"_35":269,"_23":24,"_166":-5,"_26":270},{"_42":255,"_31":256,"_29":257,"_160":-5,"_40":41,"_39":258,"_162":24,"_37":259,"_13":260,"_33":34,"_35":261,"_23":24,"_166":-5,"_26":262},{"_42":247,"_31":248,"_29":249,"_160":-5,"_40":41,"_39":250,"_162":24,"_37":251,"_13":252,"_33":34,"_35":253,"_23":24,"_166":-5,"_26":254},{"_42":239,"_31":240,"_29":241,"_160":-5,"_40":41,"_39":242,"_162":24,"_37":243,"_13":244,"_33":34,"_35":245,"_23":24,"_166":-5,"_26":246},{"_42":232,"_31":233,"_29":234,"_160":-5,"_40":41,"_39":235,"_162":58,"_37":236,"_13":213,"_33":34,"_35":237,"_23":24,"_166":-5,"_26":238},{"_42":224,"_31":225,"_29":226,"_160":-5,"_40":41,"_39":227,"_162":24,"_37":228,"_13":229,"_33":34,"_35":230,"_23":24,"_166":-5,"_26":231},{"_42":216,"_31":217,"_29":218,"_160":-5,"_40":41,"_39":219,"_162":58,"_37":220,"_13":221,"_33":34,"_35":222,"_23":24,"_166":-5,"_26":223},{"_42":208,"_31":209,"_29":210,"_160":-5,"_40":41,"_39":211,"_162":58,"_37":212,"_13":213,"_33":34,"_35":214,"_23":24,"_166":-5,"_26":215},{"_42":200,"_31":201,"_29":202,"_160":-5,"_40":41,"_39":203,"_162":24,"_37":204,"_13":205,"_33":34,"_35":206,"_23":24,"_166":-5,"_26":207},{"_42":192,"_31":193,"_29":194,"_160":-5,"_40":41,"_39":195,"_162":58,"_37":196,"_13":197,"_33":34,"_35":198,"_23":24,"_166":-5,"_26":199},{"_42":184,"_31":185,"_29":186,"_160":-5,"_40":41,"_39":187,"_162":58,"_37":188,"_13":189,"_33":34,"_35":190,"_23":24,"_166":-5,"_26":191},{"_42":176,"_31":177,"_29":56,"_160":-5,"_40":41,"_39":178,"_162":24,"_37":179,"_13":180,"_33":181,"_35":182,"_23":24,"_166":-5,"_26":183},{"_42":168,"_31":169,"_29":170,"_160":-5,"_40":41,"_39":171,"_162":24,"_37":172,"_13":173,"_33":34,"_35":174,"_23":24,"_166":-5,"_26":175},{"_42":157,"_31":158,"_29":159,"_160":-5,"_40":41,"_39":161,"_162":24,"_37":163,"_13":164,"_33":34,"_35":165,"_23":24,"_166":-5,"_26":167},"72a438d2-5d6a-11f1-b70c-1607ee4eb77e","types.ts","151549be-56ca-11f1-bc41-ee650bb23af1","parentId",639,"latest",["D",1780284719626],15,"export interface Integration {\n id: string;\n name: string; // 'ahrefs'|'ga4'|'gsc'|'linkedin'|'outlook'|'wordpress'|'youtube'|'producthunt'\n api_key?: string;\n api_secret?: string;\n access_token?: string;\n refresh_token?: string;\n extra_config?: string; // JSON string\n connected_at?: string;\n updated_at?: string;\n}\n\nexport interface LLMConfig {\n id: string;\n // QA fix: gemini was already routed in callLLM but missing from the union, causing TS noise.\n provider: 'anthropic' | 'openai' | 'gemini' | 'custom';\n api_key: string;\n endpoint?: string;\n model: string;\n updated_at?: string;\n // Per-provider key presence flags (computed from the 4 api_key_ columns) —\n // used by Settings UI to render \"(saved)\" pills without leaking the actual key.\n has_key?: { anthropic: boolean; openai: boolean; gemini: boolean; custom: boolean };\n}\n\nexport interface Agent {\n id: string;\n name: string;\n workspace: 'marketing' | 'sales';\n description: string;\n system_prompt: string;\n tools: string; // JSON array string\n model?: string;\n enabled: number; // 0 or 1\n created_at: string;\n updated_at: string;\n created_by: 'user' | 'ai';\n // B2: budget cap (USD/month) + running spend, joined from agent_budgets\n monthly_cap?: number;\n current_spend?: number;\n}\n\nexport interface Conversation {\n id: string;\n agent_id?: string;\n workspace: string;\n messages: string; // JSON array string\n created_at: string;\n updated_at: string;\n}\n\nexport interface Task {\n id: string;\n agent_id?: string;\n workspace: string;\n title: string;\n status: 'pending' | 'running' | 'done' | 'failed';\n input?: string;\n output?: string;\n created_at: string;\n completed_at?: string;\n}\n\nexport interface ChatMessage {\n role: 'user' | 'assistant' | 'system' | 'tool';\n content: string;\n timestamp?: string;\n // Tool-use fields (OpenAI-style) carried at runtime by chat.ts / llm.ts.\n tool_calls?: any[];\n tool_call_id?: string;\n reasoning_content?: string;\n}\n","label",[158],"9d2770b4-5d6b-11f1-959f-1607ee4eb77e","router.ts","30479b74-56ca-11f1-a296-ee650bb23af1",641,["D",1780285220417],94,"import { sqlite, initDB, DEFAULT_AGENT_SEEDS } from \"./db.ts\";\nimport { listAgents, createAgent, updateAgent, deleteAgent } from \"./api/agents.ts\";\nimport { getLLMConfig, saveLLMConfig } from \"./api/llm.ts\";\nimport { handleChat, handleChatStream, listConversations, getConversationById, deleteConversation } from \"./api/chat.ts\";\nimport { handleIntegrationProxy, getCredentials, GOOGLE_SERVICES, MICROSOFT_SCOPES } from \"./api/integrations.ts\";\nimport { getAppJS } from \"./frontend_js.ts\";\nimport { getUIJS } from \"./frontend_ui.ts\";\nimport { getChartsJS } from \"./frontend_charts.ts\";\nimport { getRenderersJS } from \"./frontend_renderers.ts\";\nimport { getRenderers2JS } from \"./frontend_renderers2.ts\";\nimport { getCRMJS } from \"./frontend_crm.ts\";\nimport { getPagesJS } from \"./frontend_pages.ts\";\nimport { getCommandJS } from \"./frontend_command.ts\";\nimport { getIntelligenceJS } from \"./frontend_intelligence.ts\";\nimport { getSettingsJS } from \"./frontend_settings.ts\";\nimport { getWidgetsJS } from \"./frontend_widgets.ts\";\nimport { getObjectivesJS } from \"./frontend_objectives.ts\";\nimport { getCalendarJS } from \"./frontend_calendar.ts\";\nimport { getExperimentsJS } from \"./frontend_experiments.ts\";\nimport { getFunnelSummary, getCorrelations, getRecommendations, getLeadScores, getAIOKpis, getIntelligenceFeed } from \"./api/intelligence.ts\";\nimport { listSignals, recordSignal, actionSignal, getHotLeads, getHotLeadsCount } from \"./api/signals.ts\";\nimport { listObjectives, createObjective, updateObjective, deleteObjective, createKeyResult, updateKeyResult, getOKRDashboard, syncObjectivesFromFunnel } from \"./api/okrs.ts\";\nimport { getNSM, setNSM, recordNSMValue, getNSMProgress } from \"./api/nsm.ts\";\nimport { listCohorts, createCohort, deleteCohort, getCohortSummary, getEngagementVelocity, getAttributionSummary, recordConversion } from \"./api/aio.ts\";\nimport { getLatestDigest, generateDigest, getDigestHistory } from \"./api/digest.ts\";\nimport { listExperiments, createExperiment, updateExperiment, deleteExperiment, runExperiment, completeExperiment, getExperimentStats, suggestICE } from \"./api/experiments.ts\";\nimport { getUnifiedCalendar, getCalendarSummary, createCalendarEntry, updateCalendarEntry, deleteCalendarEntry } from \"./api/calendar_unified.ts\";\nimport { getGraphify } from \"./api/graphify.ts\";\nimport { listTasks, createTask, updateTask, deleteTask } from \"./api/tasks.ts\";\nimport { listDeals, getDeal, createDeal, updateDeal, deleteDeal, syncDealsFromLeads, listCandidates, importDeals } from \"./api/deals.ts\";\nimport { listKnowledgeFiles, getKnowledgeFile, createKnowledgeFile, updateKnowledgeFile, deleteKnowledgeFile, summarizeKnowledgeFile, buildKnowledgeGraph, AI_FEATURES } from \"./api/knowledge.ts\";\nimport { getCompanyProfile, setCompanyProfile, getBrandContext, setBrandContext } from \"./api/company.ts\";\nimport { listMemory, addMemory, updateMemory, deleteMemory } from \"./api/memory.ts\";\nimport { getCampaignsJS } from \"./frontend_campaigns.ts\";\nimport { getKnowledgeGraphJS } from \"./frontend_knowledge_graph.ts\";\nimport { getApprovalsJS } from \"./frontend_approvals.ts\";\nimport { getMethodConfig, saveMethodConfig, getMethodsMetadata } from \"./api/method_config.ts\";\nimport { listActions, enqueueAction, getAction, approveAction, rejectAction } from \"./api/actions.ts\";\nimport { processPendingRuns } from \"./cron.ts\";\nimport { getExecutedWorkCounts } from \"./api/propagate.ts\";\n\nconst corsHeaders = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, PUT, DELETE, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n};\n\nfunction json(data: unknown, status = 200): Response {\n return new Response(JSON.stringify(data), {\n status,\n headers: {\n \"Content-Type\": \"application/json\",\n ...corsHeaders,\n },\n });\n}\n\n// Mask an integration row before returning it to the client. NO endpoint (list OR upsert)\n// may return secrets in cleartext — only presence flags + a short masked hint.\nfunction maskIntegration(obj: any, googleConnected = false) {\n return {\n ...obj,\n connected: !!(obj.api_key || obj.access_token) ||\n (GOOGLE_SERVICES.includes(String(obj.name)) && googleConnected),\n has_api_key: !!obj.api_key,\n has_api_secret: !!obj.api_secret,\n has_access_token: !!obj.access_token,\n api_key: obj.api_key ? String(obj.api_key).slice(0, 4) + \"***\" : undefined,\n api_secret: obj.api_secret ? \"***\" : undefined,\n access_token: obj.access_token ? \"***\" : undefined,\n refresh_token: obj.refresh_token ? \"***\" : undefined,\n };\n}\n\n// --- OAuth CSRF state (cookie-bound) -----------------------------------------\n// On the auth-start request we mint a random state, echo it in the provider URL, and\n// store it in an HttpOnly cookie. The callback must present a matching state or we reject —\n// this prevents login-CSRF / token fixation into the shared single-tenant credential row.\nfunction newOAuthState(): { state: string; setCookie: string } {\n const state = crypto.randomUUID();\n return { state, setCookie: `oauth_state=${state}; Path=/; HttpOnly; SameSite=Lax; Max-Age=600` };\n}\nconst CLEAR_OAUTH_STATE = \"oauth_state=; Path=/; HttpOnly; SameSite=Lax; Max-Age=0\";\nfunction readCookie(req: Request, name: string): string {\n const raw = req.headers.get(\"cookie\") || \"\";\n const m = raw.match(new RegExp(\"(?:^|;\\\\s*)\" + name + \"=([^;]+)\"));\n return m ? m[1] : \"\";\n}\nfunction redirectWithCookie(location: string, setCookie?: string): Response {\n const headers: Record = { Location: location };\n if (setCookie) headers[\"Set-Cookie\"] = setCookie;\n return new Response(null, { status: 302, headers });\n}\nfunction oauthStateValid(req: Request, url: URL): boolean {\n const expected = readCookie(req, \"oauth_state\");\n const got = url.searchParams.get(\"state\") || \"\";\n return !!expected && !!got && expected === got;\n}\n\nexport async function handleRequest(req: Request, url: URL): Promise {\n const { pathname } = url;\n const method = req.method;\n\n // CORS preflight\n if (method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: corsHeaders });\n }\n\n try {\n // GET /api/health\n if (pathname === \"/api/health\" && method === \"GET\") {\n return json({ status: \"ok\", timestamp: new Date().toISOString() });\n }\n\n // GET /api/health/auth - gated token verifier. Reaching this (past the main.ts gate)\n // means the supplied APP_TOKEN is valid; the login screen uses it to confirm the password.\n if (pathname === \"/api/health/auth\" && method === \"GET\") {\n return json({ ok: true });\n }\n\n // GET /api/config - LLM runtime config + ALL app_config key/values (workspace_prefs, ROI inputs, …)\n if (pathname === \"/api/config\" && method === \"GET\") {\n const llmConfig = await getLLMConfig();\n const merged: Record = {\n provider: llmConfig.provider,\n model: llmConfig.model,\n has_key: (llmConfig as any).has_key,\n };\n try {\n const r = await sqlite.execute(\"SELECT key, value FROM app_config\");\n for (const row of (r.rows || []) as any[]) {\n const k = String(row[0]);\n let v: any = row[1];\n if (typeof v === \"string\") { try { v = JSON.parse(v); } catch { /* keep raw string */ } }\n merged[k] = v;\n }\n } catch { /* app_config missing → just return LLM fields */ }\n return json(merged);\n }\n // POST /api/config - upsert arbitrary key/value into app_config (workspace prefs, ROI inputs, temperature…).\n // Without this, Settings toggles/ROI inputs silently failed to persist (POST was 404).\n if (pathname === \"/api/config\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const updated: Record = {};\n try {\n for (const [k, v] of Object.entries(body)) {\n if (k === \"provider\" || k === \"model\" || k === \"has_key\") continue;\n const value = typeof v === \"string\" ? v : JSON.stringify(v);\n await sqlite.execute(\n \"INSERT INTO app_config (key, value) VALUES (?, ?) ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = datetime('now')\",\n [k, value],\n );\n updated[k] = v;\n }\n return json({ success: true, updated });\n } catch (e: any) {\n return json({ error: String(e?.message || e) }, 500);\n }\n }\n\n // --- Frontend JS module routes --------------------------------------------\n const jsModules: Record string> = {\n \"/api/app-js\": getAppJS,\n \"/api/app-ui\": getUIJS,\n \"/api/app-charts\": getChartsJS,\n \"/api/app-renderers\": getRenderersJS,\n \"/api/app-renderers2\": getRenderers2JS,\n \"/api/app-pages\": getPagesJS,\n \"/api/app-command\": getCommandJS,\n \"/api/app-intelligence\": getIntelligenceJS,\n \"/api/app-settings\": getSettingsJS,\n \"/api/app-widgets\": getWidgetsJS,\n \"/api/app-objectives\": getObjectivesJS,\n \"/api/app-calendar\": getCalendarJS,\n \"/api/app-experiments\": getExperimentsJS,\n \"/api/app-campaigns\": getCampaignsJS,\n \"/api/app-knowledge-graph\": getKnowledgeGraphJS,\n \"/api/app-approvals\": getApprovalsJS,\n \"/api/app-crm\": getCRMJS,\n };\n if (jsModules[pathname] && method === \"GET\") {\n return new Response(jsModules[pathname](), {\n headers: {\n \"Content-Type\": \"application/javascript\",\n \"Cache-Control\": \"no-cache, no-store, must-revalidate\",\n ...corsHeaders,\n },\n });\n }\n\n // --- Intelligence routes --------------------------------------------------\n if (pathname === \"/api/intelligence/funnel-summary\" && method === \"GET\") {\n const force = url.searchParams.get(\"force\") === \"1\";\n const days = parseInt(url.searchParams.get(\"days\") ?? \"30\", 10);\n return json(await getFunnelSummary(force, days));\n }\n if (pathname === \"/api/intelligence/correlations\" && method === \"GET\") {\n const force = url.searchParams.get(\"force\") === \"1\";\n return json(await getCorrelations(force));\n }\n if (pathname === \"/api/intelligence/recommendations\" && method === \"GET\") {\n return json(await getRecommendations());\n }\n if (pathname === \"/api/intelligence/leads\" && method === \"GET\") {\n return json(await getLeadScores());\n }\n if (pathname === \"/api/intelligence/kpis\" && method === \"GET\") {\n return json(await getAIOKpis());\n }\n if (pathname === \"/api/intelligence/feed\" && method === \"GET\") {\n return json(await getIntelligenceFeed());\n }\n // Execution-aware metric: how much real work was completed (tasks/experiments) — so Measure\n // reflects execution, not just external integration numbers.\n if (pathname === \"/api/intelligence/executed-work\" && method === \"GET\") {\n const days = parseInt(url.searchParams.get(\"days\") ?? \"30\", 10);\n return json(await getExecutedWorkCounts(days));\n }\n\n // --- Signals routes -------------------------------------------------------\n if (pathname === \"/api/signals\" && method === \"GET\") {\n const actioned = url.searchParams.get(\"actioned\");\n const lead_email = url.searchParams.get(\"lead_email\") ?? undefined;\n return json(await listSignals({ lead_email, actioned: actioned === \"true\" ? true : actioned === \"false\" ? false : undefined }));\n }\n if (pathname === \"/api/signals\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n if (!body.lead_email || !body.signal_type) {\n return json({ error: \"lead_email and signal_type are required\" }, 400);\n }\n return json(await recordSignal(body), 201);\n }\n const signalActionMatch = pathname.match(/^\\/api\\/signals\\/([^/]+)\\/action$/);\n if (signalActionMatch && method === \"PUT\") {\n const result = await actionSignal(signalActionMatch[1]);\n return result ? json(result) : json({ error: \"Signal not found\" }, 404);\n }\n if (pathname === \"/api/signals/hot-leads\" && method === \"GET\") {\n return json(await getHotLeads());\n }\n if (pathname === \"/api/signals/hot-leads/count\" && method === \"GET\") {\n return json({ count: await getHotLeadsCount() });\n }\n\n // --- OKR / Objectives routes ----------------------------------------------\n if (pathname === \"/api/okrs\" && method === \"GET\") {\n const workspace = url.searchParams.get(\"workspace\") ?? undefined;\n return json(await listObjectives({ workspace }));\n }\n if (pathname === \"/api/okrs\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createObjective(body), 201);\n }\n if (pathname === \"/api/okrs/dashboard\" && method === \"GET\") {\n return json(await getOKRDashboard());\n }\n // Explicit funnel→OKR sync (persists KR/objective progress). The GET list no longer\n // writes on read, so this is the one place that refreshes stored progress on demand.\n if (pathname === \"/api/okrs/sync\" && method === \"POST\") {\n const workspace = url.searchParams.get(\"workspace\") ?? undefined;\n return json(await syncObjectivesFromFunnel({ workspace }));\n }\n const okrIdMatch = pathname.match(/^\\/api\\/okrs\\/([^/]+)$/);\n if (okrIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n return json(await updateObjective(okrIdMatch[1], body));\n }\n if (okrIdMatch && method === \"DELETE\") {\n const ok = await deleteObjective(okrIdMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n const okrKrMatch = pathname.match(/^\\/api\\/okrs\\/([^/]+)\\/key-results$/);\n if (okrKrMatch && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createKeyResult(okrKrMatch[1], body), 201);\n }\n const krIdMatch = pathname.match(/^\\/api\\/okrs\\/key-results\\/([^/]+)$/);\n if (krIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n return json(await updateKeyResult(krIdMatch[1], body));\n }\n\n // --- NSM routes -----------------------------------------------------------\n if (pathname === \"/api/nsm\" && method === \"GET\") {\n return json(await getNSM());\n }\n if (pathname === \"/api/nsm\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await setNSM(body));\n }\n if (pathname === \"/api/nsm/record\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n await recordNSMValue(body.value);\n return json({ success: true });\n }\n if (pathname === \"/api/nsm/progress\" && method === \"GET\") {\n return json(await getNSMProgress());\n }\n\n // --- AIO / Cohort routes --------------------------------------------------\n if (pathname === \"/api/aio/cohorts\" && method === \"GET\") {\n return json(await listCohorts());\n }\n if (pathname === \"/api/aio/cohorts\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createCohort(body), 201);\n }\n const cohortIdMatch = pathname.match(/^\\/api\\/aio\\/cohorts\\/([^/]+)$/);\n if (cohortIdMatch && method === \"DELETE\") {\n return json({ success: await deleteCohort(cohortIdMatch[1]) });\n }\n if (pathname === \"/api/aio/cohort-summary\" && method === \"GET\") {\n return json(await getCohortSummary());\n }\n if (pathname === \"/api/aio/engagement-velocity\" && method === \"GET\") {\n return json(await getEngagementVelocity());\n }\n\n // --- Frontend-expected aliases (cohorts/attribution/key-results) ----------\n if (pathname === \"/api/cohorts\" && method === \"GET\") {\n return json(await getCohortSummary());\n }\n if (pathname === \"/api/cohorts\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createCohort(body), 201);\n }\n const cohortAliasMatch = pathname.match(/^\\/api\\/cohorts\\/([^/]+)$/);\n if (cohortAliasMatch && method === \"DELETE\") {\n return json({ success: await deleteCohort(cohortAliasMatch[1]) });\n }\n if (pathname === \"/api/attribution\" && method === \"GET\") {\n return json(await getAttributionSummary());\n }\n if (pathname === \"/api/attribution\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n await recordConversion(body);\n return json({ success: true }, 201);\n }\n const krAliasMatch = pathname.match(/^\\/api\\/key-results\\/([^/]+)$/);\n if (krAliasMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n return json(await updateKeyResult(krAliasMatch[1], body));\n }\n\n // --- Digest routes --------------------------------------------------------\n if (pathname === \"/api/digest\" && method === \"GET\") {\n return json(await getLatestDigest());\n }\n if (pathname === \"/api/digest/generate\" && method === \"POST\") {\n return json(await generateDigest());\n }\n if (pathname === \"/api/digest/history\" && method === \"GET\") {\n const limit = parseInt(url.searchParams.get(\"limit\") ?? \"10\", 10);\n return json(await getDigestHistory(limit));\n }\n\n // --- Experiments routes ---------------------------------------------------\n if (pathname === \"/api/experiments\" && method === \"GET\") {\n const status = url.searchParams.get(\"status\") ?? undefined;\n const funnel_stage = url.searchParams.get(\"stage\") ?? url.searchParams.get(\"funnel_stage\") ?? undefined;\n const aarrr_lever = url.searchParams.get(\"aarrr_lever\") ?? undefined;\n return json(await listExperiments({ status, funnel_stage, aarrr_lever }));\n }\n if (pathname === \"/api/experiments\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createExperiment(body), 201);\n }\n if (pathname === \"/api/experiments/stats\" && method === \"GET\") {\n return json(await getExperimentStats());\n }\n if (pathname === \"/api/experiments/ice-suggest\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await suggestICE(body));\n }\n const expActionMatch = pathname.match(/^\\/api\\/experiments\\/([^/]+)\\/(run|complete)$/);\n if (expActionMatch && method === \"POST\") {\n const id = expActionMatch[1];\n const body = await req.json().catch(() => ({}));\n const result = expActionMatch[2] === \"run\"\n ? await runExperiment(id, Number(body.baseline_value) || 0)\n : await completeExperiment(id, {\n result_value: Number(body.result_value) || 0,\n outcome: body.outcome ?? \"inconclusive\",\n learnings: body.learnings,\n });\n return result ? json(result) : json({ error: \"Experiment not found\" }, 404);\n }\n const expIdMatch = pathname.match(/^\\/api\\/experiments\\/([^/]+)$/);\n if (expIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n const updated = await updateExperiment(expIdMatch[1], body);\n return updated ? json(updated) : json({ error: \"Experiment not found\" }, 404);\n }\n if (expIdMatch && method === \"DELETE\") {\n const ok = await deleteExperiment(expIdMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- Calendar unified route -----------------------------------------------\n if (pathname === \"/api/calendar/unified\" && method === \"GET\") {\n const today = new Date().toISOString().slice(0, 10);\n const from = url.searchParams.get(\"from\") ?? url.searchParams.get(\"start\") ?? today;\n const to = url.searchParams.get(\"to\") ?? url.searchParams.get(\"end\") ?? from;\n const workspace = url.searchParams.get(\"workspace\") ?? undefined;\n const sourcesParam = url.searchParams.getAll(\"source\");\n return json(await getUnifiedCalendar({ from, to, workspace, sources: sourcesParam.length ? sourcesParam as any : undefined }));\n }\n if (pathname === \"/api/calendar/summary\" && method === \"GET\") {\n const date = url.searchParams.get(\"date\") ?? new Date().toISOString().slice(0, 10);\n return json(await getCalendarSummary(date));\n }\n if (pathname === \"/api/calendar/entries\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createCalendarEntry(body), 201);\n }\n const calEntryMatch = pathname.match(/^\\/api\\/calendar\\/entries\\/([^/]+)$/);\n if (calEntryMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n const updated = await updateCalendarEntry(calEntryMatch[1], body);\n return updated ? json(updated) : json({ error: \"Not found\" }, 404);\n }\n if (calEntryMatch && method === \"DELETE\") {\n const ok = await deleteCalendarEntry(calEntryMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- Knowledge files (AI knowledge injection + graph view) ----------------\n if (pathname === \"/api/knowledge\" && method === \"GET\") {\n return json(await listKnowledgeFiles());\n }\n if (pathname === \"/api/knowledge\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const created = await createKnowledgeFile(body);\n return created ? json(created, 201) : json({ error: \"content required\" }, 400);\n }\n if (pathname === \"/api/knowledge/graph\" && method === \"GET\") {\n return json(await buildKnowledgeGraph());\n }\n if (pathname === \"/api/knowledge/features\" && method === \"GET\") {\n return json(AI_FEATURES);\n }\n const knSummarizeMatch = pathname.match(/^\\/api\\/knowledge\\/([^/]+)\\/summarize$/);\n if (knSummarizeMatch && method === \"POST\") {\n const summary = await summarizeKnowledgeFile(knSummarizeMatch[1]);\n const file = await getKnowledgeFile(knSummarizeMatch[1]);\n return json({ summary, tokens_estimate: file?.tokens_estimate ?? 0 });\n }\n const knIdMatch = pathname.match(/^\\/api\\/knowledge\\/([^/]+)$/);\n if (knIdMatch && method === \"GET\") {\n const f = await getKnowledgeFile(knIdMatch[1]);\n return f ? json(f) : json({ error: \"Not found\" }, 404);\n }\n if (knIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n const updated = await updateKnowledgeFile(knIdMatch[1], body);\n return updated ? json(updated) : json({ error: \"Not found\" }, 404);\n }\n if (knIdMatch && method === \"DELETE\") {\n const ok = await deleteKnowledgeFile(knIdMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- Company Profile + Brand Context (Part 15.2) — the business identity ---\n if (pathname === \"/api/company-profile\" && method === \"GET\") {\n return json(await getCompanyProfile());\n }\n if (pathname === \"/api/company-profile\" && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n return json(await setCompanyProfile(body));\n }\n if (pathname === \"/api/brand-context\" && method === \"GET\") {\n return json({ text: await getBrandContext() });\n }\n if (pathname === \"/api/brand-context\" && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n await setBrandContext(String(body.text ?? \"\"));\n return json({ ok: true });\n }\n\n // --- Memory (Part 15.3) — persistent evolving business facts --------------\n if (pathname === \"/api/memory\" && method === \"GET\") {\n return json(await listMemory());\n }\n if (pathname === \"/api/memory\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const m = await addMemory(body);\n return m ? json(m, 201) : json({ error: \"content required\" }, 400);\n }\n const memIdMatch = pathname.match(/^\\/api\\/memory\\/([0-9]+)$/);\n if (memIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n const m = await updateMemory(Number(memIdMatch[1]), body);\n return m ? json(m) : json({ error: \"Not found\" }, 404);\n }\n if (memIdMatch && method === \"DELETE\") {\n const ok = await deleteMemory(Number(memIdMatch[1]));\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- Method config (prioritization methodology: ICE/RICE/WSJF) ------------\n // Backend module existed but was unrouted → the Settings methodology panel 404'd.\n if (pathname === \"/api/method-config/methods\" && method === \"GET\") {\n return json(getMethodsMetadata());\n }\n if (pathname === \"/api/method-config\" && method === \"GET\") {\n return json(await getMethodConfig());\n }\n if (pathname === \"/api/method-config\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await saveMethodConfig(body));\n }\n\n // --- Action queue (human-in-the-loop execution) ---------------------------\n if (pathname === \"/api/actions\" && method === \"GET\") {\n const status = url.searchParams.get(\"status\") ?? undefined;\n return json(await listActions(status));\n }\n if (pathname === \"/api/actions\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const created = await enqueueAction(body);\n return created ? json(created, 201) : json({ error: \"integration + endpoint required\" }, 400);\n }\n const actionApproveMatch = pathname.match(/^\\/api\\/actions\\/([^/]+)\\/approve$/);\n if (actionApproveMatch && method === \"POST\") {\n const a = await approveAction(actionApproveMatch[1]);\n return a ? json(a) : json({ error: \"Not found\" }, 404);\n }\n const actionRejectMatch = pathname.match(/^\\/api\\/actions\\/([^/]+)\\/reject$/);\n if (actionRejectMatch && method === \"POST\") {\n const a = await rejectAction(actionRejectMatch[1]);\n return a ? json(a) : json({ error: \"Not found\" }, 404);\n }\n const actionIdMatch = pathname.match(/^\\/api\\/actions\\/([^/]+)$/);\n if (actionIdMatch && method === \"GET\") {\n const a = await getAction(actionIdMatch[1]);\n return a ? json(a) : json({ error: \"Not found\" }, 404);\n }\n\n // Manual trigger for the execution engine (the interval val also calls processPendingRuns).\n if (pathname === \"/api/agents/run-pending\" && method === \"POST\") {\n const limit = parseInt(url.searchParams.get(\"limit\") ?? \"3\", 10);\n return json(await processPendingRuns(limit));\n }\n\n // --- Agent runs -----------------------------------------------------------\n const agentRunsMatch = pathname.match(/^\\/api\\/agents\\/([^/]+)\\/runs$/);\n if (agentRunsMatch && method === \"GET\") {\n const agent_id = agentRunsMatch[1];\n const limit = parseInt(url.searchParams.get(\"limit\") ?? \"20\", 10);\n const result = await sqlite.execute(\n \"SELECT * FROM agent_runs WHERE agent_id = ? ORDER BY started_at DESC LIMIT ?\",\n [agent_id, limit]\n );\n return json(result.rows.map((row) => rowToObject(result.columns, row)));\n }\n if (agentRunsMatch && method === \"POST\") {\n const agent_id = agentRunsMatch[1];\n const body = await req.json().catch(() => ({}));\n const id = crypto.randomUUID();\n const now = new Date().toISOString();\n await sqlite.execute(\n \"INSERT INTO agent_runs (id, agent_id, trigger, status, started_at) VALUES (?, ?, ?, ?, ?)\",\n [id, agent_id, body.trigger ?? \"manual\", \"pending\", now]\n );\n return json({ id, agent_id, trigger: body.trigger ?? \"manual\", status: \"pending\", started_at: now }, 201);\n }\n\n // Integrations routes\n // GET /api/integrations - list all integrations\n if (pathname === \"/api/integrations\" && method === \"GET\") {\n const result = await sqlite.execute(\"SELECT * FROM integrations ORDER BY name ASC\");\n const rows = result.rows.map((row) => rowToObject(result.columns, row));\n // Google-family services inherit the central 'google' OAuth token, so they are \"connected\"\n // whenever google is — even without their own api_key/access_token row.\n const googleConnected = rows.some((o) => o.name === \"google\" && o.access_token);\n const integrations = rows.map((obj) => maskIntegration(obj, googleConnected));\n return json(integrations);\n }\n\n // POST /api/integrations/:name - upsert integration credentials\n const integrationUpsertMatch = pathname.match(/^\\/api\\/integrations\\/([^/]+)$/);\n if (integrationUpsertMatch && method === \"POST\") {\n const name = integrationUpsertMatch[1];\n const body = await req.json().catch(() => ({}));\n const now = new Date().toISOString();\n\n const existing = await sqlite.execute(\n \"SELECT id, api_key, api_secret, access_token, refresh_token, extra_config FROM integrations WHERE name = ?\",\n [name]\n );\n\n // Preserve an existing secret when the incoming value is blank, missing, or a masked\n // placeholder (\"***\"). Saving the (masked) Settings form must NOT wipe stored credentials —\n // this is what silently cleared the Ahrefs key.\n const keepSecret = (incoming: any, current: any) => {\n if (incoming == null) return current ?? null;\n const s = String(incoming);\n if (s === \"\" || s.includes(\"***\")) return current ?? null;\n return incoming;\n };\n\n if (existing.rows.length > 0) {\n const cur = existing.rows[0] as any[]; // [id, api_key, api_secret, access_token, refresh_token, extra_config]\n await sqlite.execute(\n `UPDATE integrations SET\n api_key = ?,\n api_secret = ?,\n access_token = ?,\n refresh_token = ?,\n extra_config = ?,\n updated_at = ?\n WHERE name = ?`,\n [\n keepSecret(body.api_key, cur[1]),\n keepSecret(body.api_secret, cur[2]),\n keepSecret(body.access_token, cur[3]),\n keepSecret(body.refresh_token, cur[4]),\n body.extra_config != null ? JSON.stringify(body.extra_config) : (cur[5] ?? null),\n now,\n name,\n ]\n );\n } else {\n // Insert\n const id = crypto.randomUUID();\n await sqlite.execute(\n `INSERT INTO integrations (id, name, api_key, api_secret, access_token, refresh_token, extra_config, connected_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n id,\n name,\n body.api_key ?? null,\n body.api_secret ?? null,\n body.access_token ?? null,\n body.refresh_token ?? null,\n body.extra_config != null ? JSON.stringify(body.extra_config) : null,\n now,\n now,\n ]\n );\n }\n\n const updated = await sqlite.execute(\n \"SELECT * FROM integrations WHERE name = ?\",\n [name]\n );\n const raw = rowToObject(updated.columns, updated.rows[0]);\n // NEVER return the just-saved secrets in cleartext — mask like the list endpoint.\n const googleConnected = raw.name === \"google\"\n ? !!raw.access_token\n : !!(await getCredentials(\"google\"))?.access_token;\n return json({ success: true, integration: maskIntegration(raw, googleConnected) });\n }\n\n // DELETE /api/integrations/:name\n const integrationDeleteMatch = pathname.match(/^\\/api\\/integrations\\/([^/]+)$/);\n if (integrationDeleteMatch && method === \"DELETE\") {\n const name = integrationDeleteMatch[1];\n await sqlite.execute(\"DELETE FROM integrations WHERE name = ?\", [name]);\n return json({ success: true, message: `Integration '${name}' deleted` });\n }\n\n // GET|POST /api/integrations/:name/data - proxy to integration API\n const integrationDataMatch = pathname.match(/^\\/api\\/integrations\\/([^/]+)\\/data$/);\n if (integrationDataMatch && (method === \"GET\" || method === \"POST\")) {\n const name = integrationDataMatch[1];\n const credentials = await getCredentials(name);\n if (!credentials) {\n return json({ error: `Integration '${name}' not found or not configured` }, 404);\n }\n return handleIntegrationProxy(req, name, url.searchParams, credentials);\n }\n\n // Agents routes\n // GET /api/agents - list agents (optional ?workspace= query param)\n if (pathname === \"/api/agents\" && method === \"GET\") {\n const workspace = url.searchParams.get(\"workspace\") ?? undefined;\n const agents = await listAgents(workspace);\n return json(agents);\n }\n\n // POST /api/agents - create agent\n if (pathname === \"/api/agents\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const agent = await createAgent(body);\n return json(agent, 201);\n }\n\n // PUT /api/agents/:id - update agent\n const agentUpdateMatch = pathname.match(/^\\/api\\/agents\\/([^/]+)$/);\n if (agentUpdateMatch && method === \"PUT\") {\n const id = agentUpdateMatch[1];\n const body = await req.json().catch(() => ({}));\n const agent = await updateAgent(id, body);\n if (!agent) {\n return json({ error: `Agent '${id}' not found` }, 404);\n }\n return json(agent);\n }\n\n // DELETE /api/agents/:id - delete agent\n const agentDeleteMatch = pathname.match(/^\\/api\\/agents\\/([^/]+)$/);\n if (agentDeleteMatch && method === \"DELETE\") {\n const id = agentDeleteMatch[1];\n const deleted = await deleteAgent(id);\n if (!deleted) {\n return json({ error: `Agent '${id}' not found` }, 404);\n }\n return json({ success: true, message: `Agent '${id}' deleted` });\n }\n\n // LLM config routes\n // GET /api/llm/config - get LLM config (mask api_key)\n if (pathname === \"/api/llm/config\" && method === \"GET\") {\n const config = await getLLMConfig();\n if (!config) {\n return json({ error: \"LLM config not found\" }, 404);\n }\n // Mask api_key: show first 8 chars + ***\n const masked = {\n ...config,\n api_key: config.api_key\n ? config.api_key.slice(0, 8) + \"***\"\n : \"\",\n };\n return json(masked);\n }\n\n // POST /api/llm/config - save LLM config\n if (pathname === \"/api/llm/config\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const config = await saveLLMConfig(body);\n // Mask api_key in response\n const masked = {\n ...config,\n api_key: config.api_key\n ? config.api_key.slice(0, 8) + \"***\"\n : \"\",\n };\n return json(masked);\n }\n\n // Chat route\n // POST /api/chat - AI chat handler\n if (pathname === \"/api/chat\" && method === \"POST\") {\n return handleChat(req);\n }\n\n // POST /api/chat/stream - SSE streaming chat\n if (pathname === \"/api/chat/stream\" && method === \"POST\") {\n return handleChatStream(req);\n }\n\n // --- Conversation history routes ------------------------------------------\n if (pathname === \"/api/conversations\" && method === \"GET\") {\n return json(await listConversations());\n }\n const convIdMatch = pathname.match(/^\\/api\\/conversations\\/([^/]+)$/);\n if (convIdMatch && method === \"GET\") {\n const conv = await getConversationById(convIdMatch[1]);\n return conv ? json(conv) : json({ error: \"Not found\" }, 404);\n }\n if (convIdMatch && method === \"DELETE\") {\n const ok = await deleteConversation(convIdMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // Tasks routes\n // GET /api/tasks - list tasks (optional ?workspace= query param)\n if (pathname === \"/api/tasks\" && method === \"GET\") {\n const workspace = url.searchParams.get(\"workspace\") ?? undefined;\n const status = url.searchParams.get(\"status\") ?? undefined;\n return json(await listTasks({ workspace, status }));\n }\n\n // POST /api/tasks - create task\n if (pathname === \"/api/tasks\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createTask(body), 201);\n }\n\n // PUT /api/tasks/:id - update task (e.g. toggle status)\n const taskUpdateMatch = pathname.match(/^\\/api\\/tasks\\/([^/]+)$/);\n if (taskUpdateMatch && method === \"PUT\") {\n const id = taskUpdateMatch[1];\n const body = await req.json().catch(() => ({}));\n const updated = await updateTask(id, body);\n if (!updated) return json({ error: \"Not found\" }, 404);\n return json(updated);\n }\n\n // DELETE /api/tasks/:id - delete task\n const taskDeleteMatch = pathname.match(/^\\/api\\/tasks\\/([^/]+)$/);\n if (taskDeleteMatch && method === \"DELETE\") {\n const ok = await deleteTask(taskDeleteMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- CRM Deals (sales pipeline) ------------------------------------------\n if (pathname === \"/api/deals\" && method === \"GET\") {\n return json(await listDeals(url.searchParams.get(\"stage\") ?? undefined));\n }\n if (pathname === \"/api/deals\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await createDeal(body), 201);\n }\n // Seed deals from qualified leads (explicit; checked before the :id matcher).\n if (pathname === \"/api/deals/sync\" && method === \"POST\") {\n return json(await syncDealsFromLeads());\n }\n // Interactive import: candidate leads to review, then import the selected emails.\n if (pathname === \"/api/deals/candidates\" && method === \"GET\") {\n return json(await listCandidates());\n }\n if (pathname === \"/api/deals/import\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n return json(await importDeals(Array.isArray(body.emails) ? body.emails : []));\n }\n const dealIdMatch = pathname.match(/^\\/api\\/deals\\/([^/]+)$/);\n if (dealIdMatch && dealIdMatch[1] !== \"sync\" && method === \"GET\") {\n const d = await getDeal(dealIdMatch[1]);\n return d ? json(d) : json({ error: \"Not found\" }, 404);\n }\n if (dealIdMatch && method === \"PUT\") {\n const body = await req.json().catch(() => ({}));\n const updated = await updateDeal(dealIdMatch[1], body);\n return updated ? json(updated) : json({ error: \"Not found\" }, 404);\n }\n if (dealIdMatch && method === \"DELETE\") {\n const ok = await deleteDeal(dealIdMatch[1]);\n return ok ? json({ success: true }) : json({ error: \"Not found\" }, 404);\n }\n\n // --- Data export (CSV) ---------------------------------------------------\n if (pathname === \"/api/export/signals\" && method === \"GET\") {\n const r = await sqlite.execute(\"SELECT id, lead_email, lead_name, signal_type, signal_source, signal_strength, actioned, detected_at FROM intent_signals ORDER BY detected_at DESC\");\n const cols = r.columns;\n let csv = cols.join(\",\") + \"\\n\";\n for (const row of r.rows) {\n csv += (row as any[]).map((v) => `\"${String(v ?? \"\").replace(/\"/g, '\"\"')}\"`).join(\",\") + \"\\n\";\n }\n return new Response(csv, { headers: { \"Content-Type\": \"text/csv\", \"Content-Disposition\": 'attachment; filename=\"signals.csv\"', ...corsHeaders } });\n }\n if (pathname === \"/api/export/okrs\" && method === \"GET\") {\n const objR = await sqlite.execute(\"SELECT id, title, description, workspace, status, progress, quarter, created_at FROM objectives ORDER BY created_at DESC\");\n const krR = await sqlite.execute(\"SELECT id, objective_id, title, target_value, current_value, unit, progress, updated_at FROM key_results ORDER BY updated_at DESC\");\n let csv = \"type,\" + objR.columns.join(\",\") + \"\\n\";\n for (const row of objR.rows) {\n csv += \"objective,\" + (row as any[]).map((v) => `\"${String(v ?? \"\").replace(/\"/g, '\"\"')}\"`).join(\",\") + \"\\n\";\n }\n if (krR.rows.length > 0) {\n csv += \"\\ntype,\" + krR.columns.join(\",\") + \"\\n\";\n for (const row of krR.rows) {\n csv += \"key_result,\" + (row as any[]).map((v) => `\"${String(v ?? \"\").replace(/\"/g, '\"\"')}\"`).join(\",\") + \"\\n\";\n }\n }\n return new Response(csv, { headers: { \"Content-Type\": \"text/csv\", \"Content-Disposition\": 'attachment; filename=\"okrs.csv\"', ...corsHeaders } });\n }\n if (pathname === \"/api/export/tasks\" && method === \"GET\") {\n const r = await sqlite.execute(\"SELECT id, title, description, status, priority, workspace, due_date, created_at, updated_at FROM tasks ORDER BY created_at DESC\");\n const cols = r.columns;\n let csv = cols.join(\",\") + \"\\n\";\n for (const row of r.rows) {\n csv += (row as any[]).map((v) => `\"${String(v ?? \"\").replace(/\"/g, '\"\"')}\"`).join(\",\") + \"\\n\";\n }\n return new Response(csv, { headers: { \"Content-Type\": \"text/csv\", \"Content-Disposition\": 'attachment; filename=\"tasks.csv\"', ...corsHeaders } });\n }\n\n // --- LinkedIn OAuth flow (legacy aliases) --------------------------------\n if (pathname === \"/api/oauth/linkedin/start\" && method === \"GET\") {\n return Response.redirect(`${url.origin}/api/auth/linkedin`);\n }\n if (pathname === \"/api/oauth/linkedin/callback\" && method === \"GET\") {\n return Response.redirect(`${url.origin}/api/auth/linkedin/callback?${url.search.slice(1)}`);\n }\n\n // --- Graphify - app map for MCP / AI-native navigation -------------------\n if (pathname === \"/api/graphify\" && method === \"GET\") {\n return json(await getGraphify());\n }\n\n // --- Calendar today shortcut (TodayPanel widget) --------------------------\n if (pathname === \"/api/calendar/today\" && method === \"GET\") {\n const today = new Date().toISOString().slice(0, 10);\n return json(await getUnifiedCalendar({ from: today, to: today }));\n }\n\n // --- Digest latest alias --------------------------------------------------\n if (pathname === \"/api/digest/latest\" && method === \"GET\") {\n return json(await getLatestDigest());\n }\n\n // --- Admin purge (clear intelligence cache) -------------------------------\n if (pathname === \"/api/admin/purge\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n if (body.confirm !== \"PURGE\") return json({ error: \"Send confirm: PURGE\" }, 400);\n await sqlite.execute(\"DELETE FROM intelligence_cache\").catch(() => {});\n await sqlite.execute(\"UPDATE app_config SET value=NULL WHERE key='funnel_cache'\").catch(() => {});\n return json({ success: true, message: \"Cache purged\" });\n }\n\n // --- Remove demo/seed data ------------------------------------------------\n // The only seeded (\"demo\") rows are the starter agents. Identify them by an exact\n // match on (name, description, system_prompt) so a renamed/edited agent is never touched.\n if (pathname === \"/api/admin/seed-status\" && method === \"GET\") {\n let agents = 0;\n for (const seed of DEFAULT_AGENT_SEEDS) {\n const r = await sqlite.execute(\n \"SELECT COUNT(*) FROM agents WHERE name = ? AND description = ? AND system_prompt = ?\",\n [seed.name, seed.description, seed.system_prompt],\n ).catch(() => null);\n if (r?.rows?.length) agents += Number((r.rows[0] as any[])[0]) || 0;\n }\n return json({ ok: true, total: agents, counts: { agents } });\n }\n if (pathname === \"/api/admin/cleanup-mock\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n if (body.confirm !== \"CLEANUP\") return json({ ok: false, error: \"Send confirm: CLEANUP\" }, 400);\n let removed = 0;\n for (const seed of DEFAULT_AGENT_SEEDS) {\n const r = await sqlite.execute(\n \"DELETE FROM agents WHERE name = ? AND description = ? AND system_prompt = ?\",\n [seed.name, seed.description, seed.system_prompt],\n ).catch(() => null);\n removed += Number((r as any)?.rowsAffected ?? 0) || 0;\n }\n return json({ ok: true, total: removed });\n }\n\n // Bulk reset: wipe ALL agents (= agents + campaigns, same table) plus the related\n // run/budget/conversation rows that deleteAgent() does not cascade. Guarded by confirm.\n if (pathname === \"/api/admin/reset-agents\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n if (body.confirm !== \"CLEANUP\") return json({ ok: false, error: \"Send confirm: CLEANUP\" }, 400);\n const counts: Record = {};\n for (const table of [\"agents\", \"agent_runs\", \"agent_budgets\", \"conversations\"]) {\n const r = await sqlite.execute(`DELETE FROM ${table}`).catch(() => null);\n counts[table] = Number((r as any)?.rowsAffected ?? 0) || 0;\n }\n return json({ ok: true, counts });\n }\n\n // Factory reset for THIS app only: wipe this app's own data tables, keeping api\n // (integrations, llm_config) + knowledge (knowledge_files). IMPORTANT: val.town's\n // std/sqlite is a single ACCOUNT-WIDE database shared by all of this user's vals, so we\n // must NEVER blanket-delete every table — only the explicit list below (the tables this\n // app creates in db.ts). Guarded by confirm.\n if (pathname === \"/api/admin/reset-data\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n if (body.confirm !== \"CLEANUP\") return json({ ok: false, error: \"Send confirm: CLEANUP\" }, 400);\n const WIPE_TABLES = [\n \"agents\", \"conversations\", \"agent_budgets\", \"agent_runs\",\n \"objectives\", \"key_results\", \"north_star\", \"nsm_history\",\n \"experiments\", \"tasks\", \"tasks_legacy\", \"notes\",\n \"intent_signals\", \"growth_digests\", \"calendar_entries\", \"action_queue\",\n \"customer_cohorts\", \"lead_state_changes\", \"utm_conversions\",\n \"metric_baselines\", \"metric_history\", \"intelligence_cache\", \"app_config\",\n ];\n const counts: Record = {};\n for (const table of WIPE_TABLES) {\n const r = await sqlite.execute(`DELETE FROM ${table}`).catch(() => null);\n counts[table] = Number((r as any)?.rowsAffected ?? 0) || 0;\n }\n return json({ ok: true, kept: [\"integrations\", \"llm_config\", \"knowledge_files\"], counts });\n }\n\n // --- Google OAuth redirect flow ------------------------------------------\n if (pathname === \"/api/auth/google\" && method === \"GET\") {\n const creds = await getCredentials(\"google\");\n // OAuth client is env-driven by design (GOOGLE_CLIENT_ID/SECRET); fall back to the DB row.\n const clientId = Deno.env.get(\"GOOGLE_CLIENT_ID\") || creds?.extra_config?.client_id || creds?.extra_config?.clientId;\n const redirectUri = creds?.extra_config?.redirect_uri || creds?.extra_config?.redirectUri || `${url.origin}/api/auth/google/callback`;\n if (!clientId) {\n return new Response(\"Google OAuth not configured. Add Client ID in Settings - Google.\", { status: 400 });\n }\n const scopes = [\n \"https://www.googleapis.com/auth/userinfo.email\",\n \"https://www.googleapis.com/auth/analytics.readonly\",\n \"https://www.googleapis.com/auth/webmasters.readonly\",\n \"https://www.googleapis.com/auth/gmail.modify\",\n \"https://www.googleapis.com/auth/drive.readonly\",\n \"https://www.googleapis.com/auth/spreadsheets.readonly\",\n \"https://www.googleapis.com/auth/documents.readonly\",\n \"https://www.googleapis.com/auth/calendar\",\n \"https://www.googleapis.com/auth/youtube.readonly\",\n ].join(\" \");\n const { state, setCookie } = newOAuthState();\n const params = new URLSearchParams({\n client_id: clientId,\n redirect_uri: redirectUri,\n response_type: \"code\",\n scope: scopes,\n access_type: \"offline\",\n prompt: \"consent\",\n state,\n });\n return redirectWithCookie(`https://accounts.google.com/o/oauth2/v2/auth?${params}`, setCookie);\n }\n\n if (pathname === \"/api/auth/google/callback\" && method === \"GET\") {\n const code = url.searchParams.get(\"code\");\n const error = url.searchParams.get(\"error\");\n if (error || !code) {\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(error || \"no_code\")}`);\n }\n if (!oauthStateValid(req, url)) {\n return Response.redirect(`${url.origin}/?auth_error=state_mismatch`);\n }\n const creds = await getCredentials(\"google\");\n // Trim to avoid whitespace from copy-paste/env producing a malformed (invalid_request) token request.\n const clientId = (Deno.env.get(\"GOOGLE_CLIENT_ID\") || creds?.extra_config?.client_id || creds?.extra_config?.clientId || \"\").trim();\n const clientSecret = (Deno.env.get(\"GOOGLE_CLIENT_SECRET\") || creds?.api_secret || \"\").trim();\n const redirectUri = (creds?.extra_config?.redirect_uri || creds?.extra_config?.redirectUri || `${url.origin}/api/auth/google/callback`).trim();\n\n // Google requires non-empty client_id, client_secret, and redirect_uri for a Web-application\n // authorization_code exchange. Sending an empty required param yields error=invalid_request\n // (RFC 6749 5.2). Fail fast with a clear cause instead of a malformed request to Google.\n if (!clientId || !clientSecret) {\n const missing = !clientId && !clientSecret ? \"client_id_and_secret\"\n : !clientId ? \"client_id\" : \"client_secret\";\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(\"missing_\" + missing)}`);\n }\n\n const tokenRes = await fetch(\"https://oauth2.googleapis.com/token\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({ code, client_id: clientId, client_secret: clientSecret, redirect_uri: redirectUri, grant_type: \"authorization_code\" }),\n });\n const tokens = await tokenRes.json() as any;\n if (!tokenRes.ok || !tokens.access_token) {\n // Surface Google's error + description so invalid_request causes are diagnosable from the URL.\n const detail = [tokens.error, tokens.error_description].filter(Boolean).join(\": \") || \"token_exchange_failed\";\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(detail)}`);\n }\n\n const now = new Date().toISOString();\n const newExtraConfig = { ...(creds?.extra_config || {}), token_expiry: Date.now() + (tokens.expires_in || 3600) * 1000 };\n const existing = await sqlite.execute(\"SELECT id FROM integrations WHERE name = ?\", [\"google\"]);\n if (existing.rows.length > 0) {\n await sqlite.execute(\n \"UPDATE integrations SET access_token=?, refresh_token=?, extra_config=?, updated_at=? WHERE name='google'\",\n [tokens.access_token, tokens.refresh_token || creds?.refresh_token || null, JSON.stringify(newExtraConfig), now]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO integrations (id, name, access_token, refresh_token, extra_config, connected_at, updated_at) VALUES (?,?,?,?,?,?,?)\",\n [crypto.randomUUID(), \"google\", tokens.access_token, tokens.refresh_token || null, JSON.stringify(newExtraConfig), now, now]\n );\n }\n return redirectWithCookie(`${url.origin}/?auth=google_ok`, CLEAR_OAUTH_STATE);\n }\n\n // --- LinkedIn OAuth flow --------------------------------------------------\n if (pathname === \"/api/auth/linkedin\" && method === \"GET\") {\n const creds = await getCredentials(\"linkedin\");\n const clientId = creds?.extra_config?.clientId || creds?.extra_config?.client_id;\n const redirectUri = `${url.origin}/api/auth/linkedin/callback`;\n if (!clientId) {\n return new Response(\"LinkedIn OAuth not configured. Add Client ID in Settings - LinkedIn.\", { status: 400 });\n }\n const { state, setCookie } = newOAuthState();\n const params = new URLSearchParams({\n response_type: \"code\",\n client_id: clientId,\n redirect_uri: redirectUri,\n scope: \"openid profile email w_member_social r_basicprofile\",\n state,\n });\n return redirectWithCookie(`https://www.linkedin.com/oauth/v2/authorization?${params}`, setCookie);\n }\n\n if (pathname === \"/api/auth/linkedin/callback\" && method === \"GET\") {\n const code = url.searchParams.get(\"code\");\n const error = url.searchParams.get(\"error\");\n if (error || !code) {\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(error || \"no_code\")}`);\n }\n if (!oauthStateValid(req, url)) {\n return Response.redirect(`${url.origin}/?auth_error=state_mismatch`);\n }\n const creds = await getCredentials(\"linkedin\");\n const clientId = creds?.extra_config?.clientId || creds?.extra_config?.client_id || \"\";\n const clientSecret = creds?.api_secret || \"\";\n const redirectUri = `${url.origin}/api/auth/linkedin/callback`;\n\n const tokenRes = await fetch(\"https://www.linkedin.com/oauth/v2/accessToken\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({ grant_type: \"authorization_code\", code, redirect_uri: redirectUri, client_id: clientId, client_secret: clientSecret }),\n });\n const tokens = await tokenRes.json() as any;\n if (!tokenRes.ok || !tokens.access_token) {\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(tokens.error || \"linkedin_token_failed\")}`);\n }\n\n // Fetch LinkedIn profile to get the author URN needed for posts endpoint.\n let authorId = creds?.extra_config?.author_id || \"\";\n try {\n const profileRes = await fetch(\"https://api.linkedin.com/v2/me\", {\n headers: { Authorization: `Bearer ${tokens.access_token}` },\n });\n if (profileRes.ok) {\n const profile = await profileRes.json() as any;\n if (profile.id) authorId = `urn:li:person:${profile.id}`;\n }\n } catch { /* non-fatal - author_id can be set manually later */ }\n\n const now = new Date().toISOString();\n const newExtraConfig = { ...(creds?.extra_config || {}), token_expiry: Date.now() + (tokens.expires_in || 5184000) * 1000, author_id: authorId };\n const existing = await sqlite.execute(\"SELECT id FROM integrations WHERE name = ?\", [\"linkedin\"]);\n if (existing.rows.length > 0) {\n await sqlite.execute(\n \"UPDATE integrations SET access_token=?, extra_config=?, updated_at=? WHERE name='linkedin'\",\n [tokens.access_token, JSON.stringify(newExtraConfig), now]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO integrations (id, name, access_token, extra_config, connected_at, updated_at) VALUES (?,?,?,?,?,?)\",\n [crypto.randomUUID(), \"linkedin\", tokens.access_token, JSON.stringify(newExtraConfig), now, now]\n );\n }\n return redirectWithCookie(`${url.origin}/?auth=linkedin_ok`, CLEAR_OAUTH_STATE);\n }\n\n // --- Microsoft (Outlook) OAuth flow --------------------------------------\n // Env-driven client (MICROSOFT_CLIENT_ID/SECRET), `common` tenant. Mirrors the Google flow.\n if (pathname === \"/api/auth/microsoft\" && method === \"GET\") {\n // Client id/secret are env-first (MICROSOFT_CLIENT_ID/SECRET), falling back to the\n // `outlook` integration row (extra_config.client_id + api_secret) — same pattern as LinkedIn.\n const mcreds = await getCredentials(\"outlook\");\n const clientId = (Deno.env.get(\"MICROSOFT_CLIENT_ID\") || mcreds?.extra_config?.client_id || mcreds?.extra_config?.clientId || \"\").trim();\n if (!clientId) {\n return new Response(\"Microsoft OAuth not configured. Set MICROSOFT_CLIENT_ID (env) or add the Client ID in Settings → Outlook.\", { status: 400 });\n }\n const redirectUri = `${url.origin}/api/auth/microsoft/callback`;\n const scopes = MICROSOFT_SCOPES;\n const { state, setCookie } = newOAuthState();\n const params = new URLSearchParams({\n client_id: clientId,\n response_type: \"code\",\n redirect_uri: redirectUri,\n response_mode: \"query\",\n scope: scopes,\n state,\n });\n return redirectWithCookie(`https://login.microsoftonline.com/common/oauth2/v2.0/authorize?${params}`, setCookie);\n }\n\n if (pathname === \"/api/auth/microsoft/callback\" && method === \"GET\") {\n const code = url.searchParams.get(\"code\");\n const error = url.searchParams.get(\"error\");\n if (error || !code) {\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(error || \"no_code\")}`);\n }\n if (!oauthStateValid(req, url)) {\n return Response.redirect(`${url.origin}/?auth_error=state_mismatch`);\n }\n const creds = await getCredentials(\"outlook\");\n const clientId = (Deno.env.get(\"MICROSOFT_CLIENT_ID\") || creds?.extra_config?.client_id || creds?.extra_config?.clientId || \"\").trim();\n const clientSecret = (Deno.env.get(\"MICROSOFT_CLIENT_SECRET\") || creds?.api_secret || \"\").trim();\n const redirectUri = `${url.origin}/api/auth/microsoft/callback`;\n if (!clientId || !clientSecret) {\n const missing = !clientId && !clientSecret ? \"client_id_and_secret\" : !clientId ? \"client_id\" : \"client_secret\";\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(\"missing_microsoft_\" + missing)}`);\n }\n const scopes = MICROSOFT_SCOPES;\n const tokenRes = await fetch(\"https://login.microsoftonline.com/common/oauth2/v2.0/token\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({ client_id: clientId, client_secret: clientSecret, code, redirect_uri: redirectUri, grant_type: \"authorization_code\", scope: scopes }),\n });\n const tokens = await tokenRes.json() as any;\n if (!tokenRes.ok || !tokens.access_token) {\n const detail = [tokens.error, tokens.error_description].filter(Boolean).join(\": \") || \"token_exchange_failed\";\n return Response.redirect(`${url.origin}/?auth_error=${encodeURIComponent(detail)}`);\n }\n\n const now = new Date().toISOString();\n const newExtraConfig = { ...(creds?.extra_config || {}), token_expiry: Date.now() + (tokens.expires_in || 3600) * 1000 };\n const existing = await sqlite.execute(\"SELECT id FROM integrations WHERE name = ?\", [\"outlook\"]);\n if (existing.rows.length > 0) {\n await sqlite.execute(\n \"UPDATE integrations SET access_token=?, refresh_token=?, extra_config=?, updated_at=? WHERE name='outlook'\",\n [tokens.access_token, tokens.refresh_token || creds?.refresh_token || null, JSON.stringify(newExtraConfig), now]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO integrations (id, name, access_token, refresh_token, extra_config, connected_at, updated_at) VALUES (?,?,?,?,?,?,?)\",\n [crypto.randomUUID(), \"outlook\", tokens.access_token, tokens.refresh_token || null, JSON.stringify(newExtraConfig), now, now]\n );\n }\n return redirectWithCookie(`${url.origin}/?auth=outlook_ok`, CLEAR_OAUTH_STATE);\n }\n\n // --- Notes routes (TeamHub collaborative notes) ---------------------------\n if (pathname === \"/api/notes\" && method === \"GET\") {\n const workspace = url.searchParams.get(\"workspace\") ?? \"marketing\";\n const result = await sqlite.execute(\n \"SELECT * FROM notes WHERE workspace=? ORDER BY pinned DESC, created_at DESC\",\n [workspace]\n );\n return json(result.rows.map((r) => rowToObject(result.columns, r)));\n }\n if (pathname === \"/api/notes\" && method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const id = crypto.randomUUID();\n const now = new Date().toISOString();\n await sqlite.execute(\n \"INSERT INTO notes (id, content, pinned, workspace, created_at, updated_at) VALUES (?,?,?,?,?,?)\",\n [id, body.content ?? \"\", body.pinned ? 1 : 0, body.workspace ?? \"marketing\", now, now]\n );\n return json({ id, content: body.content ?? \"\", pinned: !!body.pinned, workspace: body.workspace ?? \"marketing\", created_at: now, updated_at: now }, 201);\n }\n const noteMatch = pathname.match(/^\\/api\\/notes\\/([^/]+)$/);\n if (noteMatch && method === \"PUT\") {\n const id = noteMatch[1];\n const body = await req.json().catch(() => ({}));\n const now = new Date().toISOString();\n const setParts: string[] = [\"updated_at=?\"];\n const setArgs: any[] = [now];\n if (body.pinned !== undefined) { setParts.push(\"pinned=?\"); setArgs.push(body.pinned ? 1 : 0); }\n if (body.content !== undefined) { setParts.push(\"content=?\"); setArgs.push(body.content); }\n await sqlite.execute(\n `UPDATE notes SET ${setParts.join(\", \")} WHERE id=?`,\n [...setArgs, id]\n );\n const r = await sqlite.execute(\"SELECT * FROM notes WHERE id=?\", [id]);\n if (!r.rows.length) return json({ error: \"Not found\" }, 404);\n return json(rowToObject(r.columns, r.rows[0]));\n }\n if (noteMatch && method === \"DELETE\") {\n await sqlite.execute(\"DELETE FROM notes WHERE id=?\", [noteMatch[1]]);\n return json({ success: true });\n }\n\n // No route matched\n return json({ error: \"Not found\", path: pathname }, 404);\n } catch (err) {\n const message = err instanceof Error ? err.message : String(err);\n console.error(`[router] Error handling ${method} ${pathname}:`, message);\n return json({ error: \"Internal server error\", detail: message }, 500);\n }\n}\n\n/**\n * Converts a sqlite result row (array of values) to a plain object\n * using the columns array from the result set.\n */\nfunction rowToObject(columns: string[], row: unknown[]): Record {\n const obj: Record = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n return obj;\n}\n",[169],"9d9db260-5d6b-11f1-bf4a-1607ee4eb77e","main.ts",642,["D",1780285221175],13,"http","import { initDB } from \"./db.ts\";\nimport { handleRequest } from \"./router.ts\";\nimport { getFrontendHTML } from \"./frontend.ts\";\nimport { seedKnowledge } from \"./api/knowledge_seed.ts\";\n\nlet initialized = false;\n\n// Shared-secret gate. Set APP_TOKEN in the val's environment. Every /api/* route\n// requires `Authorization: Bearer ` (or `?token=` for download links),\n// EXCEPT the public paths below. Fail-closed: if APP_TOKEN is unset, data routes 503.\nconst APP_TOKEN = (Deno.env.get(\"APP_TOKEN\") || \"\").trim();\n\n// Login verification temporarily disabled — flip to `false` to re-enable the gate.\nconst AUTH_DISABLED = true;\n\n// Public /api paths that must stay reachable without the app token:\n// - /api/app-* : the frontend's own JS modules (static code, no secrets; loaded by \n \n \n \n \n \n \n \n \n\n\n

\n \n\n`;\n}\n",[322],"9c936de2-5d6b-11f1-a71c-1607ee4eb77e","db.ts","2f68e1b8-56ca-11f1-b821-ee650bb23af1",640,["D",1780285219474],36,"import { sqlite as _sqlite } from \"https://esm.town/v/std/sqlite\";\n\n// Compatibility shim: support sqlite.execute(sql, args) two-arg form\n// Val.town's libSQL client only accepts execute(string) or execute({sql, args})\nconst _exec = _sqlite.execute.bind(_sqlite);\nexport const sqlite = {\n ..._sqlite,\n execute: async (stmt: string | { sql: string; args?: any[] }, args?: any[]): Promise => {\n if (typeof stmt === \"string\" && args !== undefined) {\n return _exec({ sql: stmt, args });\n }\n return _exec(stmt as any);\n },\n};\n\nexport async function initDB() {\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS integrations (\n id TEXT PRIMARY KEY,\n name TEXT NOT NULL UNIQUE,\n api_key TEXT,\n api_secret TEXT,\n access_token TEXT,\n refresh_token TEXT,\n extra_config TEXT,\n connected_at TEXT,\n updated_at TEXT\n )\n `);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS llm_config (\n id TEXT PRIMARY KEY DEFAULT 'default',\n provider TEXT NOT NULL DEFAULT 'anthropic',\n api_key TEXT NOT NULL DEFAULT '',\n endpoint TEXT,\n model TEXT NOT NULL DEFAULT 'claude-sonnet-4-6',\n updated_at TEXT\n )\n `);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS agents (\n id TEXT PRIMARY KEY,\n name TEXT NOT NULL,\n workspace TEXT NOT NULL,\n description TEXT,\n system_prompt TEXT,\n tools TEXT DEFAULT '[]',\n model TEXT,\n enabled INTEGER DEFAULT 1,\n created_at TEXT NOT NULL,\n updated_at TEXT NOT NULL,\n created_by TEXT DEFAULT 'user'\n )\n `);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS conversations (\n id TEXT PRIMARY KEY,\n agent_id TEXT,\n workspace TEXT NOT NULL,\n messages TEXT DEFAULT '[]',\n created_at TEXT NOT NULL,\n updated_at TEXT NOT NULL\n )\n `);\n\n // F4 migration: the original tasks table (TEXT id, agent_id, input/output) is replaced by\n // the framework-layer tasks schema (INTEGER id, ICE fields, OKR/experiment links). If a\n // legacy table exists, rename it so the new CREATE further down can build the right shape.\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(tasks)\");\n const have = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n // Old schema is identifiable by `agent_id` + `input` + no `priority`. New schema has `priority` + `ice_score`.\n const isLegacy = have.has(\"agent_id\") && have.has(\"input\") && !have.has(\"ice_score\");\n if (isLegacy) {\n await sqlite.execute(\"ALTER TABLE tasks RENAME TO tasks_legacy\");\n }\n } catch { /* table doesn't exist yet — fine, new CREATE will run */ }\n\n // Co-working: shared notes for team collaboration on strategies.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS notes (\n id TEXT PRIMARY KEY,\n author TEXT,\n content TEXT NOT NULL,\n workspace TEXT,\n pinned INTEGER DEFAULT 0,\n created_at TEXT NOT NULL,\n updated_at TEXT\n )\n `);\n\n // Intelligence layer: per-metric 30d baselines used for health scoring.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS metric_baselines (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n integration TEXT NOT NULL,\n metric_key TEXT NOT NULL,\n baseline_value REAL,\n baseline_period TEXT DEFAULT '30d',\n last_updated TEXT DEFAULT (datetime('now')),\n UNIQUE(integration, metric_key)\n )\n `);\n\n // Intelligence cache — funnel summary, correlations, recommendations, LLM-generated feed.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS intelligence_cache (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n cache_key TEXT UNIQUE NOT NULL,\n payload TEXT NOT NULL,\n generated_at TEXT DEFAULT (datetime('now')),\n expires_at TEXT NOT NULL\n )\n `);\n\n // App-wide key/value config: ROI inputs (avg_deal_size, est_marketing_spend),\n // workspace preferences (ultra_density, translucency, hardware_accel, etc.).\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS app_config (\n key TEXT PRIMARY KEY,\n value TEXT,\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n\n // Metric history — rolling timeseries for delta_7d / delta_30d computations.\n // Pruned to last 60 days on insert to keep size bounded.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS metric_history (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n integration TEXT NOT NULL,\n metric_key TEXT NOT NULL,\n value REAL NOT NULL,\n recorded_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_metric_history_key ON metric_history(integration, metric_key, recorded_at DESC)`);\n\n // A5: rolling baseline columns — added on existing metric_baselines tables via\n // PRAGMA-guarded ALTER so we never fail on re-run.\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(metric_baselines)\");\n const have = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (!have.has(\"rolling_avg\")) {\n await sqlite.execute(\"ALTER TABLE metric_baselines ADD COLUMN rolling_avg REAL\");\n }\n if (!have.has(\"sample_count\")) {\n await sqlite.execute(\"ALTER TABLE metric_baselines ADD COLUMN sample_count INTEGER DEFAULT 0\");\n }\n } catch { /* non-critical: warm-start baselines still work */ }\n\n // QA fix: per-provider key columns on llm_config so switching providers doesn't\n // overwrite the previous provider's key. PRAGMA-guarded ALTER (re-run safe).\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(llm_config)\");\n const have = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n for (const col of [\"api_key_anthropic\", \"api_key_openai\", \"api_key_gemini\", \"api_key_custom\"]) {\n if (!have.has(col)) await sqlite.execute(`ALTER TABLE llm_config ADD COLUMN ${col} TEXT DEFAULT ''`);\n }\n if (!have.has(\"endpoint_custom\")) {\n await sqlite.execute(\"ALTER TABLE llm_config ADD COLUMN endpoint_custom TEXT DEFAULT ''\");\n }\n } catch { /* non-critical: legacy single-key still works */ }\n\n // Agent budget tracking — real monthly LLM-cost cap per agent.\n // current_spend incremented on each LLM call, reset at period_start rollover.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS agent_budgets (\n agent_id TEXT PRIMARY KEY,\n monthly_cap REAL DEFAULT 250,\n current_spend REAL DEFAULT 0,\n period_start TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n\n // C3: agent execution log. One row per run (manual or scheduled). Latest 50 power the\n // AgentCard timeline. status: queued → running → completed | failed | cancelled.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS agent_runs (\n id TEXT PRIMARY KEY,\n agent_id TEXT NOT NULL,\n status TEXT NOT NULL,\n trigger TEXT DEFAULT 'manual',\n started_at TEXT DEFAULT (datetime('now')),\n finished_at TEXT,\n duration_ms INTEGER,\n input TEXT,\n output TEXT,\n error TEXT\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_agent_runs_agent ON agent_runs(agent_id, started_at DESC)`);\n\n // D1: customer cohorts. One row per cohort (e.g. signup month). Powers cohort-survival LTV\n // beyond the simple avg_deal × retention_months heuristic. retention_curve is a JSON array\n // [m1, m2, m3 ...] of retained customer counts. arpu = average revenue per user (monthly).\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS customer_cohorts (\n id TEXT PRIMARY KEY,\n cohort_label TEXT NOT NULL,\n cohort_size INTEGER NOT NULL,\n arpu REAL NOT NULL,\n retention_curve TEXT NOT NULL,\n created_at TEXT DEFAULT (datetime('now'))\n )\n `);\n\n // D3: lead state changes — every MQL→SQL transition gets a row so we can compute\n // engagement velocity (avg days lead spent in each stage).\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS lead_state_changes (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n lead_email TEXT NOT NULL,\n from_state TEXT,\n to_state TEXT NOT NULL,\n changed_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_lead_state_email ON lead_state_changes(lead_email, changed_at)`);\n\n // D5: utm_conversions — channel attribution log. Populated when a conversion fires\n // (manually pushed via POST /api/attribution or scraped from GA4 in a later phase).\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS utm_conversions (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n utm_source TEXT,\n utm_medium TEXT,\n utm_campaign TEXT,\n conversion_value REAL DEFAULT 0,\n converted_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_utm_source ON utm_conversions(utm_source, converted_at DESC)`);\n\n // ─── FRAMEWORK LAYER (F1-F4): NSM, OKRs, Experiments, Tasks ─────────────────\n\n // F1: North Star Metric (singleton — only one active row; previous deleted on set)\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS north_star (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n metric_name TEXT NOT NULL,\n metric_source TEXT NOT NULL,\n metric_key TEXT,\n current_value REAL,\n target_value REAL NOT NULL,\n target_date TEXT NOT NULL,\n unit TEXT DEFAULT 'count',\n description TEXT,\n set_at TEXT DEFAULT (datetime('now')),\n set_by TEXT DEFAULT 'admin'\n )\n `);\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS nsm_history (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n nsm_id INTEGER REFERENCES north_star(id) ON DELETE CASCADE,\n value REAL NOT NULL,\n recorded_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_nsm_history_id ON nsm_history(nsm_id, recorded_at DESC)`);\n\n // F2: OKRs (Objectives + Key Results)\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS objectives (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n title TEXT NOT NULL,\n description TEXT,\n quarter TEXT NOT NULL,\n workspace TEXT DEFAULT 'marketing',\n owner TEXT,\n status TEXT DEFAULT 'active',\n progress REAL DEFAULT 0,\n created_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS key_results (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n objective_id INTEGER NOT NULL REFERENCES objectives(id) ON DELETE CASCADE,\n title TEXT NOT NULL,\n metric_source TEXT,\n metric_key TEXT,\n start_value REAL DEFAULT 0,\n current_value REAL DEFAULT 0,\n target_value REAL NOT NULL,\n unit TEXT DEFAULT 'count',\n confidence REAL DEFAULT 0.5,\n progress REAL DEFAULT 0,\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_kr_objective ON key_results(objective_id)`);\n\n // F3: ICE Experiments (ice_score + uplift_pct computed in app code — sqlite GENERATED\n // column support is uneven; safer to project them client-side from impact/confidence/ease).\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS experiments (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n title TEXT NOT NULL,\n hypothesis TEXT,\n objective_id INTEGER REFERENCES objectives(id) ON DELETE SET NULL,\n funnel_stage TEXT,\n aarrr_lever TEXT,\n status TEXT DEFAULT 'idea',\n impact REAL DEFAULT 5,\n confidence REAL DEFAULT 5,\n ease REAL DEFAULT 5,\n owner TEXT,\n start_date TEXT,\n end_date TEXT,\n channel TEXT,\n success_metric TEXT,\n baseline_value REAL,\n result_value REAL,\n outcome TEXT,\n learnings TEXT,\n created_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_exp_status ON experiments(status, updated_at DESC)`);\n\n // F4: Tasks (linked to OKRs / KRs / experiments — at least one for governance).\n // NOTE: the gap-closure prompt also referred to a \"tasks\" table — the older one (string id,\n // status enum 'pending|running|done|failed') is replaced by this richer schema. The two\n // were never wired to a shared UI so the swap is safe.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS tasks (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n title TEXT NOT NULL,\n description TEXT,\n status TEXT DEFAULT 'todo',\n priority TEXT DEFAULT 'medium',\n workspace TEXT DEFAULT 'marketing',\n assignee TEXT,\n due_date TEXT,\n objective_id INTEGER REFERENCES objectives(id) ON DELETE SET NULL,\n key_result_id INTEGER REFERENCES key_results(id) ON DELETE SET NULL,\n experiment_id INTEGER REFERENCES experiments(id) ON DELETE SET NULL,\n ice_score REAL,\n impact REAL,\n confidence REAL,\n ease REAL,\n tags TEXT,\n created_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now')),\n completed_at TEXT\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_tasks_status ON tasks(status, due_date)`);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_tasks_links ON tasks(experiment_id, key_result_id, objective_id)`);\n\n // ─── INTELLIGENCE LAYER EXTENSIONS (I1, I4): Signals + Digests ──────────────\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS intent_signals (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n lead_email TEXT NOT NULL,\n lead_name TEXT,\n company TEXT,\n signal_type TEXT NOT NULL,\n signal_source TEXT NOT NULL,\n signal_strength INTEGER DEFAULT 1,\n raw_data TEXT,\n detected_at TEXT DEFAULT (datetime('now')),\n actioned INTEGER DEFAULT 0,\n actioned_at TEXT\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_signals_email ON intent_signals(lead_email, detected_at DESC)`);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_signals_actioned ON intent_signals(actioned, detected_at DESC)`);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS growth_digests (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n week_of TEXT NOT NULL,\n payload TEXT NOT NULL,\n nsm_value_at_generation REAL,\n generated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_digest_week ON growth_digests(week_of DESC)`);\n\n // ─── UNIFIED CALENDAR (CAL1): standalone custom calendar entries ────────────\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS calendar_entries (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n title TEXT NOT NULL,\n description TEXT,\n event_date TEXT NOT NULL,\n event_time TEXT,\n end_date TEXT,\n end_time TEXT,\n event_type TEXT DEFAULT 'custom',\n workspace TEXT DEFAULT 'marketing',\n color TEXT DEFAULT 'blue',\n all_day INTEGER DEFAULT 0,\n recurrence TEXT,\n created_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_calendar_date ON calendar_entries(event_date)`);\n\n // Knowledge files — growth playbooks/frameworks/formulas injected into AI features (see api/knowledge.ts).\n // This table was missing from initDB, so the knowledge UI + /api/knowledge endpoints failed (\"no such table\").\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS knowledge_files (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n name TEXT NOT NULL,\n kind TEXT DEFAULT 'playbook',\n content TEXT NOT NULL,\n summary TEXT,\n targets TEXT DEFAULT '[]',\n source TEXT DEFAULT 'user',\n tokens_estimate INTEGER DEFAULT 0,\n enabled INTEGER DEFAULT 1,\n uploaded_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_knowledge_enabled ON knowledge_files(enabled, uploaded_at DESC)`);\n\n // Action queue — human-in-the-loop execution. The AI/agents PREPARE external actions\n // (send email, publish post, post to LinkedIn, create calendar event) here; nothing is\n // sent until a human approves, at which point the integration call runs and the result\n // is recorded. status: prepared → approved → executing → done | failed | rejected.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS action_queue (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n source_type TEXT DEFAULT 'chat',\n source_id TEXT,\n integration TEXT NOT NULL,\n endpoint TEXT NOT NULL,\n payload TEXT NOT NULL,\n preview TEXT,\n title TEXT,\n workspace TEXT DEFAULT 'marketing',\n status TEXT DEFAULT 'prepared',\n result TEXT,\n task_id TEXT,\n created_at TEXT DEFAULT (datetime('now')),\n approved_at TEXT,\n executed_at TEXT\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_action_queue_status ON action_queue(status, created_at DESC)`);\n\n // CRM: sales pipeline deals. Hybrid model — rows are auto-seeded from qualified leads\n // (getLeadScores) via POST /api/deals/sync AND can be created/edited/dragged by the user.\n // stage ∈ lead | qualified | proposal | negotiation | won | lost. lead_email links a deal back\n // to its source lead so the detail view can show that contact's intent_signals as activity.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS deals (\n id TEXT PRIMARY KEY,\n title TEXT,\n company TEXT,\n contact_name TEXT,\n contact_email TEXT,\n stage TEXT NOT NULL DEFAULT 'lead',\n amount REAL DEFAULT 0,\n currency TEXT DEFAULT 'USD',\n probability INTEGER DEFAULT 0,\n owner TEXT,\n close_date TEXT,\n source TEXT DEFAULT 'manual',\n lead_email TEXT,\n notes TEXT,\n created_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now')),\n last_activity_at TEXT\n )\n `);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_deals_stage ON deals(stage, updated_at DESC)`);\n await sqlite.execute(`CREATE INDEX IF NOT EXISTS idx_deals_lead ON deals(lead_email)`);\n\n // I3: agents.role column for Listener/Topic/Creator/Analyst/General roles (ALTER guarded)\n try {\n const ac = await sqlite.execute(\"PRAGMA table_info(agents)\");\n const aHave = new Set((ac.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (!aHave.has(\"role\")) await sqlite.execute(\"ALTER TABLE agents ADD COLUMN role TEXT DEFAULT 'general'\");\n } catch { /* non-critical */ }\n\n // Part 15.3: persistent business memory (evolving learned facts). Additive on v551.\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS memory (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n type TEXT DEFAULT 'fact',\n content TEXT NOT NULL,\n source TEXT DEFAULT 'manual',\n confidence REAL DEFAULT 0.7,\n pinned INTEGER DEFAULT 0,\n created_at TEXT DEFAULT (datetime('now')),\n updated_at TEXT DEFAULT (datetime('now'))\n )\n `);\n\n // Seed default LLM config if not exists\n const existing = await sqlite.execute(\"SELECT id FROM llm_config WHERE id = 'default'\");\n if (existing.rows.length === 0) {\n await sqlite.execute(\n \"INSERT INTO llm_config (id, provider, api_key, model, updated_at) VALUES ('default', 'anthropic', '', 'claude-sonnet-4-6', ?)\",\n [new Date().toISOString()]\n );\n }\n\n // Demo-agent seeding disabled — starting from a clean agents table.\n // Re-enable by uncommenting the call below.\n // await seedDefaultAgents();\n}\n\n// The starter \"demo\" agents seeded on first boot. Exported so the admin \"Remove demo data\"\n// route can identify + remove untouched seed rows (matched by name+description+system_prompt,\n// so a user-renamed/edited agent is never deleted).\nexport const DEFAULT_AGENT_SEEDS = [\n {\n name: \"SEO Analyst\",\n workspace: \"marketing\",\n description: \"Analyzes SEO performance using Ahrefs and Google Search Console data\",\n system_prompt: \"You are an expert SEO analyst. Analyze search performance data, identify opportunities, and provide actionable recommendations to improve organic search rankings.\",\n tools: JSON.stringify([\"ahrefs\", \"gsc\", \"ga4\"]),\n model: \"claude-sonnet-4-6\",\n },\n {\n name: \"Content Creator\",\n workspace: \"marketing\",\n description: \"Creates and manages content for WordPress, YouTube, and social media\",\n system_prompt: \"You are a creative content strategist. Help plan, write, and optimize content for blogs, videos, and social media platforms to drive engagement and conversions.\",\n tools: JSON.stringify([\"wordpress\", \"youtube\", \"ga4\"]),\n model: \"claude-sonnet-4-6\",\n },\n {\n name: \"Analytics Reporter\",\n workspace: \"marketing\",\n description: \"Generates reports from GA4 and Search Console data\",\n system_prompt: \"You are a data analyst specializing in digital marketing analytics. Create insightful reports from GA4 and Search Console data to track KPIs and identify growth opportunities.\",\n tools: JSON.stringify([\"ga4\", \"gsc\", \"ahrefs\"]),\n model: \"claude-sonnet-4-6\",\n },\n {\n name: \"LinkedIn Outreach\",\n workspace: \"sales\",\n description: \"Manages LinkedIn prospecting and connection campaigns\",\n system_prompt: \"You are a sales development expert specializing in LinkedIn outreach. Help craft personalized connection requests, follow-up messages, and engagement strategies to build pipeline.\",\n tools: JSON.stringify([\"linkedin\", \"outlook\"]),\n model: \"claude-sonnet-4-6\",\n },\n {\n name: \"Lead Researcher\",\n workspace: \"sales\",\n description: \"Researches prospects using LinkedIn and web data\",\n system_prompt: \"You are a research specialist who finds and qualifies sales leads. Analyze LinkedIn profiles, company data, and online presence to identify ideal prospects and gather intelligence for sales outreach.\",\n tools: JSON.stringify([\"linkedin\"]),\n model: \"claude-sonnet-4-6\",\n },\n {\n name: \"Email Composer\",\n workspace: \"sales\",\n description: \"Drafts and sends personalized sales emails via Outlook\",\n system_prompt: \"You are a sales email expert who crafts compelling, personalized outreach emails. Write emails that are concise, value-focused, and drive response rates while maintaining brand voice.\",\n tools: JSON.stringify([\"outlook\", \"linkedin\"]),\n model: \"claude-sonnet-4-6\",\n },\n];\n\nasync function seedDefaultAgents() {\n const existing = await sqlite.execute(\"SELECT id FROM agents LIMIT 1\");\n if (existing.rows.length > 0) return;\n\n const now = new Date().toISOString();\n for (const agent of DEFAULT_AGENT_SEEDS) {\n await sqlite.execute(\n `INSERT INTO agents (id, name, workspace, description, system_prompt, tools, model, enabled, created_at, updated_at, created_by)\n VALUES (?, ?, ?, ?, ?, ?, ?, 1, ?, ?, ?)`,\n [crypto.randomUUID(), agent.name, agent.workspace, agent.description, agent.system_prompt, agent.tools, agent.model, now, now, \"user\"]\n );\n }\n}\n\n",[330],"fb1075ca-5b1a-11f1-b807-ee650bb23af1","cron_entry.ts","fb10767e-5b1a-11f1-b807-ee650bb23af1",476,["D",1780030686427],"// Scheduled entrypoint — deploy this file as a Val Town INTERVAL/CRON trigger (separate from\n// the HTTP val in main.ts). A single file can't be both the HTTP handler AND the scheduled\n// trigger, which is why the execution engine never ran on its own before: cron.ts had a default\n// export but nothing scheduled it.\n//\n// What it does each tick:\n// 1) processPendingRuns() — runs the LLM tool-loop for any agent_runs with status='pending'\n// (external sends still go to the human-approval action_queue; nothing auto-sends).\n// 2) syncObjectivesFromFunnel() — refreshes + persists KR/objective progress from live funnel\n// data (the GET /api/okrs list no longer writes on read, so this keeps stored progress fresh).\n//\n// Schedule: set the interval in the Val Town dashboard (Triggers → e.g. every 5 minutes).\nimport { initDB } from \"./db.ts\";\nimport { processPendingRuns } from \"./cron.ts\";\nimport { syncObjectivesFromFunnel } from \"./api/okrs.ts\";\n\nexport default async function (): Promise {\n await initDB();\n const runs = await processPendingRuns(5);\n const okr = await syncObjectivesFromFunnel();\n console.log(\n `[cron] runs processed=${runs.processed} failed=${runs.failed}; ` +\n `okr synced objectives=${okr.objectives} key_results=${okr.key_results}`,\n );\n}\n",[338],"2c0d34e2-5d69-11f1-b581-1607ee4eb77e","cron.ts","50e655da-5a69-11f1-ab8c-ee650bb23af1",624,["D",1780284171703],"// Execution engine — the missing piece that makes agents actually RUN.\n// Deploy as a val.town interval val (default export runs on a schedule, e.g. every 1–5 min),\n// or trigger manually via POST /api/agents/run-pending (wired in router.ts).\n//\n// It polls agent_runs for status='pending', loads the agent config, and runs an LLM\n// tool-loop reusing chat.ts's buildTools()/executeTool(). External sends still go to the\n// human-approval action_queue (the tools enqueue, they don't send) — so the engine is safe.\nimport { sqlite } from \"./db.ts\";\nimport { getLLMConfig, callLLM } from \"./api/llm.ts\";\nimport { getAgent } from \"./api/agents.ts\";\nimport { buildTools, executeTool } from \"./api/chat.ts\";\n\nconst ENGINE_SYSTEM_SUFFIX =\n \"\\n\\nYou can ACT using tools. External sends (email / LinkedIn / WordPress) are QUEUED for human \" +\n \"approval — prepare them and they will appear in Approvals; they will NOT send until a human approves. \" +\n \"Create tasks and schedule calendar events directly. Take the single next-best action toward the goal.\";\n\nexport async function processPendingRuns(limit = 3): Promise<{ processed: number; failed: number }> {\n let processed = 0;\n let failed = 0;\n let rows: any[] = [];\n try {\n const r = await sqlite.execute(\n \"SELECT id, agent_id, input FROM agent_runs WHERE status = 'pending' ORDER BY started_at ASC LIMIT ?\",\n [limit],\n );\n rows = (r.rows || []) as any[];\n } catch {\n return { processed, failed };\n }\n\n for (const row of rows) {\n const runId = String(row[0]);\n const agentId = String(row[1]);\n const input = row[2] ? String(row[2]) : \"\";\n await sqlite.execute(\"UPDATE agent_runs SET status='running' WHERE id=?\", [runId]).catch(() => {});\n try {\n const agent = await getAgent(agentId);\n const llmConfig = await getLLMConfig();\n if (!(llmConfig as any).api_key) throw new Error(\"No LLM API key configured\");\n\n const systemPrompt =\n (agent?.system_prompt || \"You are an autonomous marketing & sales agent.\") + ENGINE_SYSTEM_SUFFIX;\n const messages: any[] = [{\n role: \"user\",\n content: input || `Run agent \"${agent?.name || agentId}\". Decide and prepare the next best action toward its goal.`,\n }];\n\n const tools = buildTools();\n let llm = await callLLM(llmConfig, messages, { systemPrompt, tools, maxTokens: 2048 });\n const actionLog: string[] = [];\n let rounds = 0;\n while (llm.tool_calls && llm.tool_calls.length > 0 && rounds < 4) {\n rounds++;\n messages.push({ role: \"assistant\", content: llm.content || \"\", tool_calls: llm.tool_calls });\n for (const tc of llm.tool_calls) {\n let args: Record = {};\n try {\n args = typeof tc.function.arguments === \"string\"\n ? JSON.parse(tc.function.arguments)\n : (tc.function.arguments || {});\n } catch { args = {}; }\n const res = await executeTool(tc.function.name, args, llmConfig);\n actionLog.push(`${tc.function.name}: ${res}`);\n messages.push({ role: \"tool\", tool_call_id: tc.id, content: res });\n }\n llm = await callLLM(llmConfig, messages, { systemPrompt, tools, maxTokens: 2048 });\n }\n\n const output = ((llm.content || \"\") + (actionLog.length ? \"\\n\\n[actions]\\n\" + actionLog.join(\"\\n\") : \"\")).slice(0, 8000);\n await sqlite.execute(\n \"UPDATE agent_runs SET status='completed', finished_at=datetime('now'), output=? WHERE id=?\",\n [output, runId],\n );\n processed++;\n } catch (e: any) {\n failed++;\n await sqlite.execute(\n \"UPDATE agent_runs SET status='failed', finished_at=datetime('now'), error=? WHERE id=?\",\n [String(e?.message || e), runId],\n ).catch(() => {});\n }\n }\n return { processed, failed };\n}\n\n// val.town interval-val entrypoint.\nexport default async function (): Promise {\n const result = await processPendingRuns(3);\n console.log(`[cron] agent runs processed=${result.processed} failed=${result.failed}`);\n}\n",[345],"eb52f0ae-5d6d-11f1-95b0-1607ee4eb77e","skills","00cb9425-5deb-432c-b10f-0fcd3d61109e",644,["D",1780286210534],"directory",[352],"d0f43d42-5b4e-11f1-8d86-ee650bb23af1","renderers","d81e0fb0-4134-482f-8b60-863ef03f710e",520,["D",1780052949625],[359],"156bfa70-56ca-11f1-8a9a-ee650bb23af1","api","e712dcc1-c69e-4e42-a401-18299fcf7843",["D",1779556136787],[365],"pulls","branches","forks","envVars",{"_376":378,"_379":62},"connections",{"_376":276,"_377":62},"count","unfilledCount",4,"emptyCount",{"_42":364,"_31":365,"_29":366,"_160":-5,"_40":41,"_39":189,"_162":58,"_37":1037,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_472":1038,"_26":1039},{"_42":1032,"_31":917,"_29":911,"_160":366,"_40":41,"_39":1033,"_162":58,"_37":1034,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_472":1035,"_26":1036},{"_42":1024,"_31":1025,"_29":1026,"_160":911,"_40":41,"_39":1027,"_162":24,"_37":1028,"_13":815,"_33":34,"_35":1029,"_23":24,"_166":-5,"_472":1030,"_26":1031},{"_42":1017,"_31":688,"_29":1018,"_160":911,"_40":41,"_39":1019,"_162":24,"_37":1020,"_13":62,"_33":34,"_35":1021,"_23":24,"_166":-5,"_472":1022,"_26":1023},{"_42":1010,"_31":680,"_29":1011,"_160":911,"_40":41,"_39":1012,"_162":24,"_37":1013,"_13":62,"_33":34,"_35":1014,"_23":24,"_166":-5,"_472":1015,"_26":1016},{"_42":1003,"_31":672,"_29":1004,"_160":911,"_40":41,"_39":1005,"_162":24,"_37":1006,"_13":189,"_33":34,"_35":1007,"_23":24,"_166":-5,"_472":1008,"_26":1009},{"_42":996,"_31":664,"_29":997,"_160":911,"_40":41,"_39":998,"_162":24,"_37":999,"_13":378,"_33":34,"_35":1000,"_23":24,"_166":-5,"_472":1001,"_26":1002},{"_42":989,"_31":656,"_29":990,"_160":911,"_40":41,"_39":991,"_162":58,"_37":992,"_13":62,"_33":34,"_35":993,"_23":24,"_166":-5,"_472":994,"_26":995},{"_42":982,"_31":648,"_29":983,"_160":911,"_40":41,"_39":984,"_162":24,"_37":985,"_13":189,"_33":34,"_35":986,"_23":24,"_166":-5,"_472":987,"_26":988},{"_42":975,"_31":640,"_29":976,"_160":911,"_40":41,"_39":977,"_162":24,"_37":978,"_13":62,"_33":34,"_35":979,"_23":24,"_166":-5,"_472":980,"_26":981},{"_42":968,"_31":632,"_29":969,"_160":911,"_40":41,"_39":970,"_162":24,"_37":971,"_13":189,"_33":34,"_35":972,"_23":24,"_166":-5,"_472":973,"_26":974},{"_42":960,"_31":624,"_29":961,"_160":911,"_40":41,"_39":962,"_162":24,"_37":963,"_13":964,"_33":34,"_35":965,"_23":24,"_166":-5,"_472":966,"_26":967},{"_42":953,"_31":616,"_29":954,"_160":911,"_40":41,"_39":955,"_162":24,"_37":956,"_13":378,"_33":34,"_35":957,"_23":24,"_166":-5,"_472":958,"_26":959},{"_42":946,"_31":608,"_29":947,"_160":911,"_40":41,"_39":948,"_162":24,"_37":949,"_13":268,"_33":34,"_35":950,"_23":24,"_166":-5,"_472":951,"_26":952},{"_42":939,"_31":600,"_29":940,"_160":911,"_40":41,"_39":941,"_162":58,"_37":942,"_13":62,"_33":34,"_35":943,"_23":24,"_166":-5,"_472":944,"_26":945},{"_42":932,"_31":592,"_29":933,"_160":911,"_40":41,"_39":934,"_162":58,"_37":935,"_13":62,"_33":34,"_35":936,"_23":24,"_166":-5,"_472":937,"_26":938},{"_42":925,"_31":584,"_29":926,"_160":911,"_40":41,"_39":927,"_162":58,"_37":928,"_13":62,"_33":34,"_35":929,"_23":24,"_166":-5,"_472":930,"_26":931},{"_42":918,"_31":576,"_29":919,"_160":911,"_40":41,"_39":920,"_162":24,"_37":921,"_13":189,"_33":34,"_35":922,"_23":24,"_166":-5,"_472":923,"_26":924},{"_42":909,"_31":568,"_29":910,"_160":911,"_40":41,"_39":912,"_162":24,"_37":913,"_13":62,"_33":34,"_35":914,"_23":24,"_166":-5,"_472":915,"_26":916},{"_42":901,"_31":902,"_29":903,"_160":366,"_40":41,"_39":904,"_162":24,"_37":905,"_13":62,"_33":34,"_35":906,"_23":24,"_166":-5,"_472":907,"_26":908},{"_42":893,"_31":894,"_29":895,"_160":366,"_40":41,"_39":896,"_162":24,"_37":897,"_13":164,"_33":34,"_35":898,"_23":24,"_166":-5,"_472":899,"_26":900},{"_42":885,"_31":886,"_29":887,"_160":366,"_40":41,"_39":888,"_162":24,"_37":889,"_13":268,"_33":34,"_35":890,"_23":24,"_166":-5,"_472":891,"_26":892},{"_42":877,"_31":878,"_29":879,"_160":366,"_40":41,"_39":880,"_162":24,"_37":881,"_13":378,"_33":34,"_35":882,"_23":24,"_166":-5,"_472":883,"_26":884},{"_42":868,"_31":869,"_29":870,"_160":366,"_40":41,"_39":871,"_162":24,"_37":872,"_13":873,"_33":34,"_35":874,"_23":24,"_166":-5,"_472":875,"_26":876},{"_42":860,"_31":861,"_29":862,"_160":366,"_40":41,"_39":863,"_162":24,"_37":864,"_13":62,"_33":34,"_35":865,"_23":24,"_166":-5,"_472":866,"_26":867},{"_42":852,"_31":853,"_29":854,"_160":366,"_40":41,"_39":855,"_162":24,"_37":856,"_13":213,"_33":34,"_35":857,"_23":24,"_166":-5,"_472":858,"_26":859},{"_42":844,"_31":845,"_29":846,"_160":366,"_40":41,"_39":847,"_162":24,"_37":848,"_13":189,"_33":34,"_35":849,"_23":24,"_166":-5,"_472":850,"_26":851},{"_42":836,"_31":837,"_29":838,"_160":366,"_40":41,"_39":839,"_162":24,"_37":840,"_13":297,"_33":34,"_35":841,"_23":24,"_166":-5,"_472":842,"_26":843},{"_42":828,"_31":829,"_29":830,"_160":366,"_40":41,"_39":831,"_162":58,"_37":832,"_13":62,"_33":34,"_35":833,"_23":24,"_166":-5,"_472":834,"_26":835},{"_42":819,"_31":820,"_29":821,"_160":366,"_40":41,"_39":822,"_162":24,"_37":823,"_13":824,"_33":34,"_35":825,"_23":24,"_166":-5,"_472":826,"_26":827},{"_42":810,"_31":811,"_29":812,"_160":366,"_40":41,"_39":813,"_162":24,"_37":814,"_13":815,"_33":34,"_35":816,"_23":24,"_166":-5,"_472":817,"_26":818},{"_42":802,"_31":803,"_29":804,"_160":366,"_40":41,"_39":805,"_162":24,"_37":806,"_13":268,"_33":34,"_35":807,"_23":24,"_166":-5,"_472":808,"_26":809},{"_42":794,"_31":795,"_29":796,"_160":366,"_40":41,"_39":797,"_162":24,"_37":798,"_13":62,"_33":34,"_35":799,"_23":24,"_166":-5,"_472":800,"_26":801},{"_42":785,"_31":786,"_29":787,"_160":366,"_40":41,"_39":788,"_162":24,"_37":789,"_13":790,"_33":34,"_35":791,"_23":24,"_166":-5,"_472":792,"_26":793},{"_42":777,"_31":778,"_29":779,"_160":366,"_40":41,"_39":780,"_162":58,"_37":781,"_13":276,"_33":34,"_35":782,"_23":24,"_166":-5,"_472":783,"_26":784},{"_42":769,"_31":770,"_29":771,"_160":366,"_40":41,"_39":772,"_162":24,"_37":773,"_13":62,"_33":34,"_35":774,"_23":24,"_166":-5,"_472":775,"_26":776},{"_42":761,"_31":762,"_29":763,"_160":366,"_40":41,"_39":764,"_162":24,"_37":765,"_13":62,"_33":34,"_35":766,"_23":24,"_166":-5,"_472":767,"_26":768},{"_42":753,"_31":754,"_29":755,"_160":366,"_40":41,"_39":756,"_162":24,"_37":757,"_13":268,"_33":34,"_35":758,"_23":24,"_166":-5,"_472":759,"_26":760},{"_42":745,"_31":746,"_29":747,"_160":366,"_40":41,"_39":748,"_162":24,"_37":749,"_13":244,"_33":34,"_35":750,"_23":24,"_166":-5,"_472":751,"_26":752},{"_42":737,"_31":738,"_29":739,"_160":366,"_40":41,"_39":740,"_162":24,"_37":741,"_13":189,"_33":34,"_35":742,"_23":24,"_166":-5,"_472":743,"_26":744},{"_42":729,"_31":730,"_29":731,"_160":366,"_40":41,"_39":732,"_162":24,"_37":733,"_13":378,"_33":34,"_35":734,"_23":24,"_166":-5,"_472":735,"_26":736},{"_42":721,"_31":722,"_29":723,"_160":366,"_40":41,"_39":724,"_162":24,"_37":725,"_13":62,"_33":34,"_35":726,"_23":24,"_166":-5,"_472":727,"_26":728},{"_42":713,"_31":714,"_29":715,"_160":366,"_40":41,"_39":716,"_162":58,"_37":717,"_13":62,"_33":34,"_35":718,"_23":24,"_166":-5,"_472":719,"_26":720},{"_42":705,"_31":706,"_29":707,"_160":366,"_40":41,"_39":708,"_162":24,"_37":709,"_13":213,"_33":34,"_35":710,"_23":24,"_166":-5,"_472":711,"_26":712},{"_42":358,"_31":359,"_29":360,"_160":-5,"_40":41,"_39":361,"_162":58,"_37":702,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_472":703,"_26":704},{"_42":695,"_31":696,"_29":697,"_160":360,"_40":41,"_39":361,"_162":58,"_37":698,"_13":62,"_33":34,"_35":699,"_23":24,"_166":-5,"_472":700,"_26":701},{"_42":687,"_31":688,"_29":689,"_160":360,"_40":41,"_39":690,"_162":58,"_37":691,"_13":62,"_33":34,"_35":692,"_23":24,"_166":-5,"_472":693,"_26":694},{"_42":679,"_31":680,"_29":681,"_160":360,"_40":41,"_39":682,"_162":58,"_37":683,"_13":62,"_33":34,"_35":684,"_23":24,"_166":-5,"_472":685,"_26":686},{"_42":671,"_31":672,"_29":673,"_160":360,"_40":41,"_39":674,"_162":58,"_37":675,"_13":276,"_33":34,"_35":676,"_23":24,"_166":-5,"_472":677,"_26":678},{"_42":663,"_31":664,"_29":665,"_160":360,"_40":41,"_39":666,"_162":58,"_37":667,"_13":62,"_33":34,"_35":668,"_23":24,"_166":-5,"_472":669,"_26":670},{"_42":655,"_31":656,"_29":657,"_160":360,"_40":41,"_39":658,"_162":58,"_37":659,"_13":62,"_33":34,"_35":660,"_23":24,"_166":-5,"_472":661,"_26":662},{"_42":647,"_31":648,"_29":649,"_160":360,"_40":41,"_39":650,"_162":24,"_37":651,"_13":62,"_33":34,"_35":652,"_23":24,"_166":-5,"_472":653,"_26":654},{"_42":639,"_31":640,"_29":641,"_160":360,"_40":41,"_39":642,"_162":58,"_37":643,"_13":62,"_33":34,"_35":644,"_23":24,"_166":-5,"_472":645,"_26":646},{"_42":631,"_31":632,"_29":633,"_160":360,"_40":41,"_39":634,"_162":24,"_37":635,"_13":62,"_33":34,"_35":636,"_23":24,"_166":-5,"_472":637,"_26":638},{"_42":623,"_31":624,"_29":625,"_160":360,"_40":41,"_39":626,"_162":58,"_37":627,"_13":62,"_33":34,"_35":628,"_23":24,"_166":-5,"_472":629,"_26":630},{"_42":615,"_31":616,"_29":617,"_160":360,"_40":41,"_39":618,"_162":58,"_37":619,"_13":62,"_33":34,"_35":620,"_23":24,"_166":-5,"_472":621,"_26":622},{"_42":607,"_31":608,"_29":609,"_160":360,"_40":41,"_39":610,"_162":58,"_37":611,"_13":62,"_33":34,"_35":612,"_23":24,"_166":-5,"_472":613,"_26":614},{"_42":599,"_31":600,"_29":601,"_160":360,"_40":41,"_39":602,"_162":58,"_37":603,"_13":62,"_33":34,"_35":604,"_23":24,"_166":-5,"_472":605,"_26":606},{"_42":591,"_31":592,"_29":593,"_160":360,"_40":41,"_39":594,"_162":58,"_37":595,"_13":276,"_33":34,"_35":596,"_23":24,"_166":-5,"_472":597,"_26":598},{"_42":583,"_31":584,"_29":585,"_160":360,"_40":41,"_39":586,"_162":58,"_37":587,"_13":62,"_33":34,"_35":588,"_23":24,"_166":-5,"_472":589,"_26":590},{"_42":575,"_31":576,"_29":577,"_160":360,"_40":41,"_39":578,"_162":58,"_37":579,"_13":62,"_33":34,"_35":580,"_23":24,"_166":-5,"_472":581,"_26":582},{"_42":567,"_31":568,"_29":569,"_160":360,"_40":41,"_39":570,"_162":24,"_37":571,"_13":62,"_33":34,"_35":572,"_23":24,"_166":-5,"_472":573,"_26":574},{"_42":351,"_31":352,"_29":353,"_160":-5,"_40":41,"_39":354,"_162":58,"_37":564,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_472":565,"_26":566},{"_42":560,"_31":552,"_29":548,"_160":353,"_40":41,"_39":354,"_162":58,"_37":561,"_13":62,"_33":356,"_35":-5,"_23":24,"_166":-5,"_472":562,"_26":563},{"_42":553,"_31":554,"_29":555,"_160":548,"_40":41,"_39":354,"_162":58,"_37":556,"_13":62,"_33":34,"_35":557,"_23":24,"_166":-5,"_472":558,"_26":559},{"_42":547,"_31":32,"_29":30,"_160":548,"_40":41,"_39":14,"_162":58,"_37":549,"_13":62,"_33":34,"_35":36,"_23":24,"_166":-5,"_472":550,"_26":551},{"_42":344,"_31":345,"_29":346,"_160":-5,"_40":41,"_39":347,"_162":24,"_37":544,"_13":189,"_33":34,"_35":349,"_23":24,"_166":-5,"_472":545,"_26":546},{"_42":337,"_31":338,"_29":339,"_160":-5,"_40":41,"_39":340,"_162":24,"_37":541,"_13":62,"_33":34,"_35":342,"_23":24,"_166":-5,"_472":542,"_26":543},{"_42":329,"_31":330,"_29":331,"_160":-5,"_40":41,"_39":332,"_162":24,"_37":538,"_13":334,"_33":34,"_35":335,"_23":24,"_166":-5,"_472":539,"_26":540},{"_42":321,"_31":322,"_29":323,"_160":-5,"_40":41,"_39":324,"_162":24,"_37":535,"_13":326,"_33":34,"_35":327,"_23":24,"_166":-5,"_472":536,"_26":537},{"_42":314,"_31":315,"_29":316,"_160":-5,"_40":41,"_39":317,"_162":24,"_37":532,"_13":62,"_33":34,"_35":319,"_23":24,"_166":-5,"_472":533,"_26":534},{"_42":307,"_31":308,"_29":309,"_160":-5,"_40":41,"_39":310,"_162":58,"_37":529,"_13":244,"_33":34,"_35":312,"_23":24,"_166":-5,"_472":530,"_26":531},{"_42":300,"_31":301,"_29":302,"_160":-5,"_40":41,"_39":303,"_162":24,"_37":526,"_13":276,"_33":34,"_35":305,"_23":24,"_166":-5,"_472":527,"_26":528},{"_42":293,"_31":294,"_29":295,"_160":-5,"_40":41,"_39":195,"_162":24,"_37":523,"_13":297,"_33":34,"_35":298,"_23":24,"_166":-5,"_472":524,"_26":525},{"_42":286,"_31":287,"_29":288,"_160":-5,"_40":41,"_39":289,"_162":24,"_37":520,"_13":244,"_33":34,"_35":291,"_23":24,"_166":-5,"_472":521,"_26":522},{"_42":279,"_31":280,"_29":281,"_160":-5,"_40":41,"_39":282,"_162":24,"_37":517,"_13":189,"_33":34,"_35":284,"_23":24,"_166":-5,"_472":518,"_26":519},{"_42":271,"_31":272,"_29":273,"_160":-5,"_40":41,"_39":274,"_162":24,"_37":514,"_13":276,"_33":34,"_35":277,"_23":24,"_166":-5,"_472":515,"_26":516},{"_42":263,"_31":264,"_29":265,"_160":-5,"_40":41,"_39":266,"_162":24,"_37":511,"_13":268,"_33":34,"_35":269,"_23":24,"_166":-5,"_472":512,"_26":513},{"_42":255,"_31":256,"_29":257,"_160":-5,"_40":41,"_39":258,"_162":24,"_37":508,"_13":260,"_33":34,"_35":261,"_23":24,"_166":-5,"_472":509,"_26":510},{"_42":247,"_31":248,"_29":249,"_160":-5,"_40":41,"_39":250,"_162":24,"_37":505,"_13":252,"_33":34,"_35":253,"_23":24,"_166":-5,"_472":506,"_26":507},{"_42":239,"_31":240,"_29":241,"_160":-5,"_40":41,"_39":242,"_162":24,"_37":502,"_13":244,"_33":34,"_35":245,"_23":24,"_166":-5,"_472":503,"_26":504},{"_42":232,"_31":233,"_29":234,"_160":-5,"_40":41,"_39":235,"_162":58,"_37":499,"_13":213,"_33":34,"_35":237,"_23":24,"_166":-5,"_472":500,"_26":501},{"_42":224,"_31":225,"_29":226,"_160":-5,"_40":41,"_39":227,"_162":24,"_37":496,"_13":229,"_33":34,"_35":230,"_23":24,"_166":-5,"_472":497,"_26":498},{"_42":216,"_31":217,"_29":218,"_160":-5,"_40":41,"_39":219,"_162":58,"_37":493,"_13":221,"_33":34,"_35":222,"_23":24,"_166":-5,"_472":494,"_26":495},{"_42":208,"_31":209,"_29":210,"_160":-5,"_40":41,"_39":211,"_162":58,"_37":490,"_13":213,"_33":34,"_35":214,"_23":24,"_166":-5,"_472":491,"_26":492},{"_42":200,"_31":201,"_29":202,"_160":-5,"_40":41,"_39":203,"_162":24,"_37":487,"_13":205,"_33":34,"_35":206,"_23":24,"_166":-5,"_472":488,"_26":489},{"_42":192,"_31":193,"_29":194,"_160":-5,"_40":41,"_39":195,"_162":58,"_37":484,"_13":197,"_33":34,"_35":198,"_23":24,"_166":-5,"_472":485,"_26":486},{"_42":184,"_31":185,"_29":186,"_160":-5,"_40":41,"_39":187,"_162":58,"_37":481,"_13":189,"_33":34,"_35":190,"_23":24,"_166":-5,"_472":482,"_26":483},{"_42":176,"_31":177,"_29":56,"_160":-5,"_40":41,"_39":178,"_162":24,"_37":478,"_13":180,"_33":181,"_35":182,"_23":24,"_166":-5,"_472":479,"_26":480},{"_42":168,"_31":169,"_29":170,"_160":-5,"_40":41,"_39":171,"_162":24,"_37":475,"_13":173,"_33":34,"_35":174,"_23":24,"_166":-5,"_472":476,"_26":477},{"_42":157,"_31":158,"_29":159,"_160":-5,"_40":41,"_39":161,"_162":24,"_37":471,"_13":164,"_33":34,"_35":165,"_23":24,"_166":-5,"_472":473,"_26":474},["D",1780284719626],"children",[],[158],["D",1780285220417],[],[169],["D",1780285221175],[],[177],["D",1780029981482],[],[185],["D",1779775316358],[],[193],["D",1780284622528],[],[201],["D",1780052948479],[],[209],["D",1780052947805],[],[217],["D",1780053918280],[],[225],["D",1780030685043],[],[233],["D",1780284184740],[],[240],["D",1780284617617],[],[248],["D",1780284182056],[],[256],["D",1779846828121],[],[264],["D",1779857291624],[],[272],["D",1780053734733],[],[280],["D",1780284613253],[],[287],["D",1779775316358],[],[294],["D",1779951334999],[],[301],["D",1780053308385],[],[308],["D",1779959471170],[],[315],["D",1779846828527],[],[322],["D",1780285219474],[],[330],["D",1780030686427],[],[338],["D",1780284171703],[],[345],"ec2a70d8-5d6d-11f1-9fc6-1607ee4eb77e","3a0403db-269a-4e55-8cb2-8c836891cb4c",["D",1780286212010],[],[352,552,32],"catalog","eb519254-5d6d-11f1-95b0-1607ee4eb77e","part_1.ts","eb519312-5d6d-11f1-95b0-1607ee4eb77e",["D",1780286210534],"// GENERATED by deploy/gen_skills_catalog.mjs — do not edit by hand.\nimport type { SkillDef } from \"../index.ts\";\nexport const SKILLS_PART_1: SkillDef[] = [\n {\"name\":\"ab-testing\",\"category\":\"CRO\",\"description\":\"When the user wants to plan, design, or implement an A/B test or experiment, or build a growth experimentation program. Also use when the user mentions \\\"A/B test,\\\" \\\"split test,\\\" \\\"experiment,\\\" \\\"test this change,\\\" \\\"variant copy,\\\" \\\"multivariate test,\\\" \\\"hypothesis,\\\" \\\"should I test this,\\\" \\\"which version is better,\\\" \\\"test two versions,\\\" \\\"statistical significance,\\\" \\\"how long should I run this test,\\\" \\\"growth experiments,\\\" \\\"experiment velocity,\\\" \\\"experiment backlog,\\\" \\\"ICE score,\\\" \\\"experimentation program,\\\" or \\\"experiment playbook.\\\" Use this whenever someone is comparing two approaches and wants to measure which performs better, or when they want to build a systematic experimentation practice. For tracking implementation, see analytics. For page-level conversion optimization, see cro.\",\"version\":\"2.0.0\",\"body\":\"# A/B Test Setup\\n\\nYou are an expert in experimentation and A/B testing. Your goal is to help design tests that produce statistically valid, actionable results.\\n\\n## Initial Assessment\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nBefore designing a test, understand:\\n\\n1. **Test Context** - What are you trying to improve? What change are you considering?\\n2. **Current State** - Baseline conversion rate? Current traffic volume?\\n3. **Constraints** - Technical complexity? Timeline? Tools available?\\n\\n---\\n\\n## Core Principles\\n\\n### 1. Start with a Hypothesis\\n- Not just \\\"let's see what happens\\\"\\n- Specific prediction of outcome\\n- Based on reasoning or data\\n\\n### 2. Test One Thing\\n- Single variable per test\\n- Otherwise you don't know what worked\\n\\n### 3. Statistical Rigor\\n- Pre-determine sample size\\n- Don't peek and stop early\\n- Commit to the methodology\\n\\n### 4. Measure What Matters\\n- Primary metric tied to business value\\n- Secondary metrics for context\\n- Guardrail metrics to prevent harm\\n\\n---\\n\\n## Hypothesis Framework\\n\\n### Structure\\n\\n```\\nBecause [observation/data],\\nwe believe [change]\\nwill cause [expected outcome]\\nfor [audience].\\nWe'll know this is true when [metrics].\\n```\\n\\n### Example\\n\\n**Weak**: \\\"Changing the button color might increase clicks.\\\"\\n\\n**Strong**: \\\"Because users report difficulty finding the CTA (per heatmaps and feedback), we believe making the button larger and using contrasting color will increase CTA clicks by 15%+ for new visitors. We'll measure click-through rate from page view to signup start.\\\"\\n\\n---\\n\\n## Test Types\\n\\n| Type | Description | Traffic Needed |\\n|------|-------------|----------------|\\n| A/B | Two versions, single change | Moderate |\\n| A/B/n | Multiple variants | Higher |\\n| MVT | Multiple changes in combinations | Very high |\\n| Split URL | Different URLs for variants | Moderate |\\n\\n---\\n\\n## Sample Size\\n\\n### Quick Reference\\n\\n| Baseline | 10% Lift | 20% Lift | 50% Lift |\\n|----------|----------|----------|----------|\\n| 1% | 150k/variant | 39k/variant | 6k/variant |\\n| 3% | 47k/variant | 12k/variant | 2k/variant |\\n| 5% | 27k/variant | 7k/variant | 1.2k/variant |\\n| 10% | 12k/variant | 3k/variant | 550/variant |\\n\\n**Calculators:**\\n- [Evan Miller's](https://www.evanmiller.org/ab-testing/sample-size.html)\\n- [Optimizely's](https://www.optimizely.com/sample-size-calculator/)\\n\\n**For detailed sample size tables and duration calculations**: See [references/sample-size-guide.md](references/sample-size-guide.md)\\n\\n---\\n\\n## Metrics Selection\\n\\n### Primary Metric\\n- Single metric that matters most\\n- Directly tied to hypothesis\\n- What you'll use to call the test\\n\\n### Secondary Metrics\\n- Support primary metric interpretation\\n- Explain why/how the change worked\\n\\n### Guardrail Metrics\\n- Things that shouldn't get worse\\n- Stop test if significantly negative\\n\\n### Example: Pricing Page Test\\n- **Primary**: Plan selection rate\\n- **Secondary**: Time on page, plan distribution\\n- **Guardrail**: Support tickets, refund rate\\n\\n---\\n\\n## Designing Variants\\n\\n### What to Vary\\n\\n| Category | Examples |\\n|----------|----------|\\n| Headlines/Copy | Message angle, value prop, specificity, tone |\\n| Visual Design | Layout, color, images, hierarchy |\\n| CTA | Button copy, size, placement, number |\\n| Content | Information included, order, amount, social proof |\\n\\n### Best Practices\\n- Single, meaningful change\\n- Bold enough to make a difference\\n- True to the hypothesis\\n\\n---\\n\\n## Traffic Allocation\\n\\n| Approach | Split | When to Use |\\n|----------|-------|-------------|\\n| Standard | 50/50 | Default for A/B |\\n| Conservative | 90/10, 80/20 | Limit risk of bad variant |\\n| Ramping | Start small, increase | Technical risk mitigation |\\n\\n**Considerations:**\\n- Consistency: Users see same variant on return\\n- Balanced exposure across time of day/week\\n\\n---\\n\\n## Implementation\\n\\n### Client-Side\\n- JavaScript modifies page after load\\n- Quick to implement, can cause flicker\\n- Tools: PostHog, Optimizely, VWO\\n\\n### Server-Side\\n- Variant determined before render\\n- No flicker, requires dev work\\n- Tools: PostHog, LaunchDarkly, Split\\n\\n---\\n\\n## Running the Test\\n\\n### Pre-Launch Checklist\\n- [ ] Hypothesis documented\\n- [ ] Primary metric defined\\n- [ ] Sample size calculated\\n- [ ] Variants implemented correctly\\n- [ ] Tracking verified\\n- [ ] QA completed on all variants\\n\\n### During the Test\\n\\n**DO:**\\n- Monitor for technical issues\\n- Check segment quality\\n- Document external factors\\n\\n**Avoid:**\\n- Peek at results and stop early\\n- Make changes to variants\\n- Add traffic from new sources\\n\\n### The Peeking Problem\\nLooking at results before reaching sample size and stopping early leads to false positives and wrong decisions. Pre-commit to sample size and trust the process.\\n\\n---\\n\\n## Analyzing Results\\n\\n### Statistical Significance\\n- 95% confidence = p-value < 0.05\\n- Means <5% chance result is random\\n- Not a guarantee—just a threshold\\n\\n### Analysis Checklist\\n\\n1. **Reach sample size?** If not, result is preliminary\\n2. **Statistically significant?** Check confidence intervals\\n3. **Effect size meaningful?** Compare to MDE, project impact\\n4. **Secondary metrics consistent?** Support the primary?\\n5. **Guardrail concerns?** Anything get worse?\\n6. **Segment differences?** Mobile vs. desktop? New vs. returning?\\n\\n### Interpreting Results\\n\\n| Result | Conclusion |\\n|--------|------------|\\n| Significant winner | Implement variant |\\n| Significant loser | Keep control, learn why |\\n| No significant difference | Need more traffic or bolder test |\\n| Mixed signals | Dig deeper, maybe segment |\\n\\n---\\n\\n## Documentation\\n\\nDocument every test with:\\n- Hypothesis\\n- Variants (with screenshots)\\n- Results (sample, metrics, significance)\\n- Decision and learnings\\n\\n**For templates**: See [references/test-templates.md](references/test-templates.md)\\n\\n---\\n\\n## Growth Experimentation Program\\n\\nIndividual tests are valuable. A continuous experimentation program is a compounding asset. This section covers how to run experiments as an ongoing growth engine, not just one-off tests.\\n\\n### The Experiment Loop\\n\\n```\\n1. Generate hypotheses (from data, research, competitors, customer feedback)\\n2. Prioritize with ICE scoring\\n3. Design and run the test\\n4. Analyze results with statistical rigor\\n5. Promote winners to a playbook\\n6. Generate new hypotheses from learnings\\n→ Repeat\\n```\\n\\n### Hypothesis Generation\\n\\nFeed your experiment backlog from multiple sources:\\n\\n| Source | What to Look For |\\n|--------|-----------------|\\n| Analytics | Drop-off points, low-converting pages, underperforming segments |\\n| Customer research | Pain points, confusion, unmet expectations |\\n| Competitor analysis | Features, messaging, or UX patterns they use that you don't |\\n| Support tickets | Recurring questions or complaints about conversion flows |\\n| Heatmaps/recordings | Where users hesitate, rage-click, or abandon |\\n| Past experiments | \\\"Significant loser\\\" tests often reveal new angles to try |\\n\\n### ICE Prioritization\\n\\nScore each hypothesis 1-10 on three dimensions:\\n\\n| Dimension | Question |\\n|-----------|----------|\\n| **Impact** | If this works, how much will it move the primary metric? |\\n| **Confidence** | How sure are we this will work? (Based on data, not gut.) |\\n| **Ease** | How fast and cheap can we ship and measure this? |\\n\\n**ICE Score** = (Impact + Confidence + Ease) / 3\\n\\nRun highest-scoring experiments first. Re-score monthly as context changes.\\n\\n### Experiment Velocity\\n\\nTrack your experimentation rate as a leading indicator of growth:\\n\\n| Metric | Target |\\n|--------|--------|\\n| Experiments launched per month | 4-8 for most teams |\\n| Win rate | 20-30% is common for mature programs (sustained higher rates may indicate conservative hypotheses) |\\n| Average test duration | 2-4 weeks |\\n| Backlog depth | 20+ hypotheses queued |\\n| Cumulative lift | Compound gains from all winners |\\n\\n### The Experiment Playbook\\n\\nWhen a test wins, don't just implement it — document the pattern:\\n\\n```\\n## [Experiment Name]\\n**Date**: [date]\\n**Hypothesis**: [the hypothesis]\\n**Sample size**: [n per variant]\\n**Result**: [winner/loser/inconclusive] — [primary metric] changed by [X%] (95% CI: [range], p=[value])\\n**Guardrails**: [any guardrail metrics and their outcomes]\\n**Segment deltas**: [notable differences by device, segment, or cohort]\\n**Why it worked/failed**: [analysis]\\n**Pattern**: [the reusable insight — e.g., \\\"social proof near pricing CTAs increases plan selection\\\"]\\n**Apply to**: [other pages/flows where this pattern might work]\\n**Status**: [implemented / parked / needs follow-up test]\\n```\\n\\nOver time, your playbook becomes a library of proven growth patterns specific to your product and audience.\\n\\n### Experiment Cadence\\n\\n**Weekly (30 min)**: Review running experiments for technical issues and guardrail metrics. Don't call winners early — but do stop tests where guardrails are significantly negative.\\n\\n**Bi-weekly**: Conclude completed experiments. Analyze results, update playbook, launch next experiment from backlog.\\n\\n**Monthly (1 hour)**: Review experiment velocity, win rate, cumulative lift. Replenish hypothesis backlog. Re-prioritize with ICE.\\n\\n**Quarterly**: Audit the playbook. Which patterns have been applied broadly? Which winning patterns haven't been scaled yet? What areas of the funnel are under-tested?\\n\\n---\\n\\n## Common Mistakes\\n\\n### Test Design\\n- Testing too small a change (undetectable)\\n- Testing too many things (can't isolate)\\n- No clear hypothesis\\n\\n### Execution\\n- Stopping early\\n- Changing things mid-test\\n- Not checking implementation\\n\\n### Analysis\\n- Ignoring confidence intervals\\n- Cherry-picking segments\\n- Over-interpreting inconclusive results\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What's your current conversion rate?\\n2. How much traffic does this page get?\\n3. What change are you considering and why?\\n4. What's the smallest improvement worth detecting?\\n5. What tools do you have for testing?\\n6. Have you tested this area before?\\n\\n---\\n\\n## Related Skills\\n\\n- **cro**: For generating test ideas based on CRO principles\\n- **analytics**: For setting up test measurement\\n- **copywriting**: For creating variant copy\"},\n {\"name\":\"ad-creative\",\"category\":\"Ads\",\"description\":\"\\\"When the user wants to generate, iterate, or scale ad creative — headlines, descriptions, primary text, or full ad variations — for any paid advertising platform. Also use when the user mentions 'ad copy variations,' 'ad creative,' 'generate headlines,' 'RSA headlines,' 'bulk ad copy,' 'ad iterations,' 'creative testing,' 'ad performance optimization,' 'write me some ads,' 'Facebook ad copy,' 'Google ad headlines,' 'LinkedIn ad text,' or 'I need more ad variations.' Use this whenever someone needs to produce ad copy at scale or iterate on existing ads. For campaign strategy and targeting, see ads. For landing page copy, see copywriting.\\\"\",\"version\":\"2.0.0\",\"body\":\"# Ad Creative\\n\\nYou are an expert performance creative strategist. Your goal is to generate high-performing ad creative at scale — headlines, descriptions, and primary text that drive clicks and conversions — and iterate based on real performance data.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nGather this context (ask if not provided):\\n\\n### 1. Platform & Format\\n- What platform? (Google Ads, Meta, LinkedIn, TikTok, Twitter/X)\\n- What ad format? (Search RSAs, display, social feed, stories, video)\\n- Are there existing ads to iterate on, or starting from scratch?\\n\\n### 2. Product & Offer\\n- What are you promoting? (Product, feature, free trial, demo, lead magnet)\\n- What's the core value proposition?\\n- What makes this different from competitors?\\n\\n### 3. Audience & Intent\\n- Who is the target audience?\\n- What stage of awareness? (Problem-aware, solution-aware, product-aware)\\n- What pain points or desires drive them?\\n\\n### 4. Performance Data (if iterating)\\n- What creative is currently running?\\n- Which headlines/descriptions are performing best? (CTR, conversion rate, ROAS)\\n- Which are underperforming?\\n- What angles or themes have been tested?\\n\\n### 5. Constraints\\n- Brand voice guidelines or words to avoid?\\n- Compliance requirements? (Industry regulations, platform policies)\\n- Any mandatory elements? (Brand name, trademark symbols, disclaimers)\\n\\n---\\n\\n## How This Skill Works\\n\\nThis skill supports two modes:\\n\\n### Mode 1: Generate from Scratch\\nWhen starting fresh, you generate a full set of ad creative based on product context, audience insights, and platform best practices.\\n\\n### Mode 2: Iterate from Performance Data\\nWhen the user provides performance data (CSV, paste, or API output), you analyze what's working, identify patterns in top performers, and generate new variations that build on winning themes while exploring new angles.\\n\\nThe core loop:\\n\\n```\\nPull performance data → Identify winning patterns → Generate new variations → Validate specs → Deliver\\n```\\n\\n---\\n\\n## Platform Specs\\n\\nPlatforms reject or truncate creative that exceeds these limits, so verify every piece of copy fits before delivering.\\n\\n### Google Ads (Responsive Search Ads)\\n\\n| Element | Limit | Quantity |\\n|---------|-------|----------|\\n| Headline | 30 characters | Up to 15 |\\n| Description | 90 characters | Up to 4 |\\n| Display URL path | 15 characters each | 2 paths |\\n\\n**RSA rules:**\\n- Headlines must make sense independently and in any combination\\n- Pin headlines to positions only when necessary (reduces optimization)\\n- Include at least one keyword-focused headline\\n- Include at least one benefit-focused headline\\n- Include at least one CTA headline\\n\\n### Meta Ads (Facebook/Instagram)\\n\\n| Element | Limit | Notes |\\n|---------|-------|-------|\\n| Primary text | 125 chars visible (up to 2,200) | Front-load the hook |\\n| Headline | 40 characters recommended | Below the image |\\n| Description | 30 characters recommended | Below headline |\\n| URL display link | 40 characters | Optional |\\n\\n### LinkedIn Ads\\n\\n| Element | Limit | Notes |\\n|---------|-------|-------|\\n| Intro text | 150 chars recommended (600 max) | Above the image |\\n| Headline | 70 chars recommended (200 max) | Below the image |\\n| Description | 100 chars recommended (300 max) | Appears in some placements |\\n\\n### TikTok Ads\\n\\n| Element | Limit | Notes |\\n|---------|-------|-------|\\n| Ad text | 80 chars recommended (100 max) | Above the video |\\n| Display name | 40 characters | Brand name |\\n\\n### Twitter/X Ads\\n\\n| Element | Limit | Notes |\\n|---------|-------|-------|\\n| Tweet text | 280 characters | The ad copy |\\n| Headline | 70 characters | Card headline |\\n| Description | 200 characters | Card description |\\n\\nFor detailed specs and format variations, see [references/platform-specs.md](references/platform-specs.md).\\n\\n---\\n\\n## Generating Ad Visuals\\n\\nFor image and video ad creative, use generative AI tools and code-based video rendering. See [references/generative-tools.md](references/generative-tools.md) for the complete guide covering:\\n\\n- **Image generation** — Nano Banana Pro (Gemini), Flux, Ideogram for static ad images\\n- **Video generation** — Veo, Kling, Runway, Sora, Seedance, Higgsfield for video ads\\n- **Voice & audio** — ElevenLabs, OpenAI TTS, Cartesia for voiceovers, cloning, multilingual\\n- **Code-based video** — Remotion for templated, data-driven video at scale\\n- **Platform image specs** — Correct dimensions for every ad placement\\n- **Cost comparison** — Pricing for 100+ ad variations across tools\\n\\n**Recommended workflow for scaled production:**\\n1. Generate hero creative with AI tools (exploratory, high-quality)\\n2. Build Remotion templates based on winning patterns\\n3. Batch produce variations with Remotion using data feeds\\n4. Iterate — AI for new angles, Remotion for scale\\n\\n---\\n\\n## Generating Ad Copy\\n\\n### Step 1: Define Your Angles\\n\\nBefore writing individual headlines, establish 3-5 distinct **angles** — different reasons someone would click. Each angle should tap into a different motivation.\\n\\n**Common angle categories:**\\n\\n| Category | Example Angle |\\n|----------|---------------|\\n| Pain point | \\\"Stop wasting time on X\\\" |\\n| Outcome | \\\"Achieve Y in Z days\\\" |\\n| Social proof | \\\"Join 10,000+ teams who...\\\" |\\n| Curiosity | \\\"The X secret top companies use\\\" |\\n| Comparison | \\\"Unlike X, we do Y\\\" |\\n| Urgency | \\\"Limited time: get X free\\\" |\\n| Identity | \\\"Built for [specific role/type]\\\" |\\n| Contrarian | \\\"Why [common practice] doesn't work\\\" |\\n\\n### Step 2: Generate Variations per Angle\\n\\nFor each angle, generate multiple variations. Vary:\\n- **Word choice** — synonyms, active vs. passive\\n- **Specificity** — numbers vs. general claims\\n- **Tone** — direct vs. question vs. command\\n- **Structure** — short punch vs. full benefit statement\\n\\n### Step 3: Validate Against Specs\\n\\nBefore delivering, check every piece of creative against the platform's character limits. Flag anything that's over and provide a trimmed alternative.\\n\\n### Step 4: Organize for Upload\\n\\nPresent creative in a structured format that maps to the ad platform's upload requirements.\\n\\n---\\n\\n## Iterating from Performance Data\\n\\nWhen the user provides performance data, follow this process:\\n\\n### Step 1: Analyze Winners\\n\\nLook at the top-performing creative (by CTR, conversion rate, or ROAS — ask which metric matters most) and identify:\\n\\n- **Winning themes** — What topics or pain points appear in top performers?\\n- **Winning structures** — Questions? Statements? Commands? Numbers?\\n- **Winning word patterns** — Specific words or phrases that recur?\\n- **Character utilization** — Are top performers shorter or longer?\\n\\n### Step 2: Analyze Losers\\n\\nLook at the worst performers and identify:\\n\\n- **Themes that fall flat** — What angles aren't resonating?\\n- **Common patterns in low performers** — Too generic? Too long? Wrong tone?\\n\\n### Step 3: Generate New Variations\\n\\nCreate new creative that:\\n- **Doubles down** on winning themes with fresh phrasing\\n- **Extends** winning angles into new variations\\n- **Tests** 1-2 new angles not yet explored\\n- **Avoids** patterns found in underperformers\\n\\n### Step 4: Document the Iteration\\n\\nTrack what was learned and what's being tested:\\n\\n```\\n## Iteration Log\\n- Round: [number]\\n- Date: [date]\\n- Top performers: [list with metrics]\\n- Winning patterns: [summary]\\n- New variations: [count] headlines, [count] descriptions\\n- New angles being tested: [list]\\n- Angles retired: [list]\\n```\\n\\n---\\n\\n## Writing Quality Standards\\n\\n### Headlines That Click\\n\\n**Strong headlines:**\\n- Specific (\\\"Cut reporting time 75%\\\") over vague (\\\"Save time\\\")\\n- Benefits (\\\"Ship code faster\\\") over features (\\\"CI/CD pipeline\\\")\\n- Active voice (\\\"Automate your reports\\\") over passive (\\\"Reports are automated\\\")\\n- Include numbers when possible (\\\"3x faster,\\\" \\\"in 5 minutes,\\\" \\\"10,000+ teams\\\")\\n\\n**Avoid:**\\n- Jargon the audience won't recognize\\n- Claims without specificity (\\\"Best,\\\" \\\"Leading,\\\" \\\"Top\\\")\\n- All caps or excessive punctuation\\n- Clickbait that the landing page can't deliver on\\n\\n### Descriptions That Convert\\n\\nDescriptions should complement headlines, not repeat them. Use descriptions to:\\n- Add proof points (numbers, testimonials, awards)\\n- Handle objections (\\\"No credit card required,\\\" \\\"Free forever for small teams\\\")\\n- Reinforce CTAs (\\\"Start your free trial today\\\")\\n- Add urgency when genuine (\\\"Limited to first 500 signups\\\")\\n\\n---\\n\\n## Output Formats\\n\\n### Standard Output\\n\\nOrganize by angle, with character counts:\\n\\n```\\n## Angle: [Pain Point — Manual Reporting]\\n\\n### Headlines (30 char max)\\n1. \\\"Stop Building Reports by Hand\\\" (29)\\n2. \\\"Automate Your Weekly Reports\\\" (28)\\n3. \\\"Reports Done in 5 Min, Not 5 Hr\\\" (31) <- OVER LIMIT, trimmed below\\n -> \\\"Reports in 5 Min, Not 5 Hrs\\\" (27)\\n\\n### Descriptions (90 char max)\\n1. \\\"Marketing teams save 10+ hours/week with automated reporting. Start free.\\\" (73)\\n2. \\\"Connect your data sources once. Get automated reports forever. No code required.\\\" (80)\\n```\\n\\n### Bulk CSV Output\\n\\nWhen generating at scale (10+ variations), offer CSV format for direct upload:\\n\\n```csv\\nheadline_1,headline_2,headline_3,description_1,description_2,platform\\n\\\"Stop Manual Reporting\\\",\\\"Automate in 5 Minutes\\\",\\\"Join 10K+ Teams\\\",\\\"Save 10+ hrs/week on reports. Start free.\\\",\\\"Connect data sources once. Reports forever.\\\",\\\"google_ads\\\"\\n```\\n\\n### Iteration Report\\n\\nWhen iterating, include a summary:\\n\\n```\\n## Performance Summary\\n- Analyzed: [X] headlines, [Y] descriptions\\n- Top performer: \\\"[headline]\\\" — [metric]: [value]\\n- Worst performer: \\\"[headline]\\\" — [metric]: [value]\\n- Pattern: [observation]\\n\\n## New Creative\\n[organized variations]\\n\\n## Recommendations\\n- [What to pause, what to scale, what to test next]\\n```\\n\\n---\\n\\n## Batch Generation Workflow\\n\\nFor large-scale creative production (Anthropic's growth team generates 100+ variations per cycle):\\n\\n### 1. Break into sub-tasks\\n- **Headline generation** — Focused on click-through\\n- **Description generation** — Focused on conversion\\n- **Primary text generation** — Focused on engagement (Meta/LinkedIn)\\n\\n### 2. Generate in waves\\n- Wave 1: Core angles (3-5 angles, 5 variations each)\\n- Wave 2: Extended variations on top 2 angles\\n- Wave 3: Wild card angles (contrarian, emotional, specific)\\n\\n### 3. Quality filter\\n- Remove anything over character limit\\n- Remove duplicates or near-duplicates\\n- Flag anything that might violate platform policies\\n- Ensure headline/description combinations make sense together\\n\\n---\\n\\n## Common Mistakes\\n\\n- **Writing headlines that only work together** — RSA headlines get combined randomly\\n- **Ignoring character limits** — Platforms truncate without warning\\n- **All variations sound the same** — Vary angles, not just word choice\\n- **No CTA headlines** — RSAs need action-oriented headlines to drive clicks; include at least 2-3\\n- **Generic descriptions** — \\\"Learn more about our solution\\\" wastes the slot\\n- **Iterating without data** — Gut feelings are less reliable than metrics\\n- **Testing too many things at once** — Change one variable per test cycle\\n- **Retiring creative too early** — Allow 1,000+ impressions before judging\\n\\n---\\n\\n## Tool Integrations\\n\\nFor pulling performance data and managing campaigns, see the [tools registry](../../tools/REGISTRY.md).\\n\\n| Platform | Pull Performance Data | Manage Campaigns | Guide |\\n|----------|:---------------------:|:----------------:|-------|\\n| **Google Ads** | `google-ads campaigns list`, `google-ads reports get` | `google-ads campaigns create` | [google-ads.md](../../tools/integrations/google-ads.md) |\\n| **Meta Ads** | `meta-ads insights get` | `meta-ads campaigns list` | [meta-ads.md](../../tools/integrations/meta-ads.md) |\\n| **LinkedIn Ads** | `linkedin-ads analytics get` | `linkedin-ads campaigns list` | [linkedin-ads.md](../../tools/integrations/linkedin-ads.md) |\\n| **TikTok Ads** | `tiktok-ads reports get` | `tiktok-ads campaigns list` | [tiktok-ads.md](../../tools/integrations/tiktok-ads.md) |\\n\\n### Workflow: Pull Data, Analyze, Generate\\n\\n```bash\\n# 1. Pull recent ad performance\\nnode tools/clis/google-ads.js reports get --type ad_performance --date-range last_30_days\\n\\n# 2. Analyze output (identify top/bottom performers)\\n# 3. Feed winning patterns into this skill\\n# 4. Generate new variations\\n# 5. Upload to platform\\n```\\n\\n---\\n\\n## Related Skills\\n\\n- **ads**: For campaign strategy, targeting, budgets, and optimization\\n- **copywriting**: For landing page copy (where ad traffic lands)\\n- **ab-testing**: For structuring creative tests with statistical rigor\\n- **marketing-psychology**: For psychological principles behind high-performing creative\\n- **copy-editing**: For polishing ad copy before launch\"},\n {\"name\":\"ads\",\"category\":\"Ads\",\"description\":\"\\\"When the user wants help with paid advertising campaigns on Google Ads, Meta (Facebook/Instagram), LinkedIn, Twitter/X, or other ad platforms. Also use when the user mentions 'PPC,' 'paid media,' 'ROAS,' 'CPA,' 'ad campaign,' 'retargeting,' 'audience targeting,' 'Google Ads,' 'Facebook ads,' 'LinkedIn ads,' 'ad budget,' 'cost per click,' 'ad spend,' or 'should I run ads.' Use this for campaign strategy, audience targeting, bidding, and optimization. For bulk ad creative generation and iteration, see ad-creative. For landing page optimization, see cro.\\\"\",\"version\":\"2.0.1\",\"body\":\"# Paid Ads\\n\\nYou are an expert performance marketer with direct access to ad platform accounts. Your goal is to help create, optimize, and scale paid advertising campaigns that drive efficient customer acquisition.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nGather this context (ask if not provided):\\n\\n### 1. Campaign Goals\\n- What's the primary objective? (Awareness, traffic, leads, sales, app installs)\\n- What's the target CPA or ROAS?\\n- What's the monthly/weekly budget?\\n- Any constraints? (Brand guidelines, compliance, geographic)\\n\\n### 2. Product & Offer\\n- What are you promoting? (Product, free trial, lead magnet, demo)\\n- What's the landing page URL?\\n- What makes this offer compelling?\\n\\n### 3. Audience\\n- Who is the ideal customer?\\n- What problem does your product solve for them?\\n- What are they searching for or interested in?\\n- Do you have existing customer data for lookalikes?\\n\\n### 4. Current State\\n- Have you run ads before? What worked/didn't?\\n- Do you have existing pixel/conversion data?\\n- What's your current funnel conversion rate?\\n\\n---\\n\\n## Platform Selection Guide\\n\\n| Platform | Best For | Use When |\\n|----------|----------|----------|\\n| **Google Ads** | High-intent search traffic | People actively search for your solution |\\n| **Meta** | Demand generation, visual products | Creating demand, strong creative assets |\\n| **LinkedIn** | B2B, decision-makers | Job title/company targeting matters, higher price points |\\n| **Twitter/X** | Tech audiences, thought leadership | Audience is active on X, timely content |\\n| **TikTok** | Younger demographics, viral creative | Audience skews 18-34, video capacity |\\n\\n---\\n\\n## Campaign Structure Best Practices\\n\\n### Account Organization\\n\\n```\\nAccount\\n├── Campaign 1: [Objective] - [Audience/Product]\\n│ ├── Ad Set 1: [Targeting variation]\\n│ │ ├── Ad 1: [Creative variation A]\\n│ │ ├── Ad 2: [Creative variation B]\\n│ │ └── Ad 3: [Creative variation C]\\n│ └── Ad Set 2: [Targeting variation]\\n└── Campaign 2...\\n```\\n\\n### Naming Conventions\\n\\n```\\n[Platform]_[Objective]_[Audience]_[Offer]_[Date]\\n\\nExamples:\\nMETA_Conv_Lookalike-Customers_FreeTrial_2024Q1\\nGOOG_Search_Brand_Demo_Ongoing\\nLI_LeadGen_CMOs-SaaS_Whitepaper_Mar24\\n```\\n\\n### Budget Allocation\\n\\n**Testing phase (first 2-4 weeks):**\\n- 70% to proven/safe campaigns\\n- 30% to testing new audiences/creative\\n\\n**Scaling phase:**\\n- Consolidate budget into winning combinations\\n- Increase budgets 20-30% at a time\\n- Wait 3-5 days between increases for algorithm learning\\n\\n---\\n\\n## Ad Copy Frameworks\\n\\n### Key Formulas\\n\\n**Problem-Agitate-Solve (PAS):**\\n> [Problem] → [Agitate the pain] → [Introduce solution] → [CTA]\\n\\n**Before-After-Bridge (BAB):**\\n> [Current painful state] → [Desired future state] → [Your product as bridge]\\n\\n**Social Proof Lead:**\\n> [Impressive stat or testimonial] → [What you do] → [CTA]\\n\\n**For detailed templates and headline formulas**: See [references/ad-copy-templates.md](references/ad-copy-templates.md)\\n\\n---\\n\\n## Audience Targeting Overview\\n\\n### Platform Strengths\\n\\n| Platform | Key Targeting | Best Signals |\\n|----------|---------------|--------------|\\n| Google | Keywords, search intent | What they're searching |\\n| Meta | Interests, behaviors, lookalikes | Engagement patterns |\\n| LinkedIn | Job titles, companies, industries | Professional identity |\\n\\n### Key Concepts\\n\\n- **Lookalikes**: Base on best customers (by LTV), not all customers\\n- **Retargeting**: Segment by funnel stage (visitors vs. cart abandoners)\\n- **Exclusions**: Exclude existing customers and recent converters — showing ads to people who already bought wastes spend\\n\\n**For detailed targeting strategies by platform**: See [references/audience-targeting.md](references/audience-targeting.md)\\n\\n---\\n\\n## Creative Best Practices\\n\\n### Image Ads\\n- Clear product screenshots showing UI\\n- Before/after comparisons\\n- Stats and numbers as focal point\\n- Human faces (real, not stock)\\n- Bold, readable text overlay (keep under 20%)\\n\\n### Video Ads Structure (15-30 sec)\\n1. Hook (0-3 sec): Pattern interrupt, question, or bold statement\\n2. Problem (3-8 sec): Relatable pain point\\n3. Solution (8-20 sec): Show product/benefit\\n4. CTA (20-30 sec): Clear next step\\n\\n**Production tips:**\\n- Captions always (85% watch without sound)\\n- Vertical for Stories/Reels, square for feed\\n- Native feel outperforms polished\\n- First 3 seconds determine if they watch\\n\\n### Creative Testing Hierarchy\\n1. Concept/angle (biggest impact)\\n2. Hook/headline\\n3. Visual style\\n4. Body copy\\n5. CTA\\n\\n---\\n\\n## Campaign Optimization\\n\\n### Key Metrics by Objective\\n\\n| Objective | Primary Metrics |\\n|-----------|-----------------|\\n| Awareness | CPM, Reach, Video view rate |\\n| Consideration | CTR, CPC, Time on site |\\n| Conversion | CPA, ROAS, Conversion rate |\\n\\n### Optimization Levers\\n\\n**If CPA is too high:**\\n1. Check landing page (is the problem post-click?)\\n2. Tighten audience targeting\\n3. Test new creative angles\\n4. Improve ad relevance/quality score\\n5. Adjust bid strategy\\n\\n**If CTR is low:**\\n- Creative isn't resonating → test new hooks/angles\\n- Audience mismatch → refine targeting\\n- Ad fatigue → refresh creative\\n\\n**If CPM is high:**\\n- Audience too narrow → expand targeting\\n- High competition → try different placements\\n- Low relevance score → improve creative fit\\n\\n### Bid Strategy Progression\\n1. Start with manual or cost caps\\n2. Gather conversion data (50+ conversions)\\n3. Switch to automated with targets based on historical data\\n4. Monitor and adjust targets based on results\\n\\n---\\n\\n## Retargeting Strategies\\n\\n### Funnel-Based Approach\\n\\n| Funnel Stage | Audience | Message | Goal |\\n|--------------|----------|---------|------|\\n| Top | Blog readers, video viewers | Educational, social proof | Move to consideration |\\n| Middle | Pricing/feature page visitors | Case studies, demos | Move to decision |\\n| Bottom | Cart abandoners, trial users | Urgency, objection handling | Convert |\\n\\n### Retargeting Windows\\n\\n| Stage | Window | Frequency Cap |\\n|-------|--------|---------------|\\n| Hot (cart/trial) | 1-7 days | Higher OK |\\n| Warm (key pages) | 7-30 days | 3-5x/week |\\n| Cold (any visit) | 30-90 days | 1-2x/week |\\n\\n### Exclusions to Set Up\\n- Existing customers (unless upsell)\\n- Recent converters (7-14 day window)\\n- Bounced visitors (<10 sec)\\n- Irrelevant pages (careers, support)\\n\\n---\\n\\n## Reporting & Analysis\\n\\n### Weekly Review\\n- Spend vs. budget pacing\\n- CPA/ROAS vs. targets\\n- Top and bottom performing ads\\n- Audience performance breakdown\\n- Frequency check (fatigue risk)\\n- Landing page conversion rate\\n\\n### Attribution Considerations\\n- Platform attribution is inflated\\n- Use UTM parameters consistently\\n- Compare platform data to GA4\\n- Look at blended CAC, not just platform CPA\\n\\n---\\n\\n## Platform Setup\\n\\nBefore launching campaigns, ensure proper tracking and account setup.\\n\\n**For complete setup checklists by platform**: See [references/platform-setup-checklists.md](references/platform-setup-checklists.md)\\n\\n**For conversion pixel installation and event setup**: See [references/conversion-tracking.md](references/conversion-tracking.md)\\n\\n### Universal Pre-Launch Checklist\\n- [ ] Conversion tracking tested with real conversion\\n- [ ] Landing page loads fast (<3 sec)\\n- [ ] Landing page mobile-friendly\\n- [ ] UTM parameters working\\n- [ ] Budget set correctly\\n- [ ] Targeting matches intended audience\\n\\n---\\n\\n## Google RSA Output Spec (mandatory when generating RSAs)\\n\\nWhen the user requests Google Ads RSAs (Responsive Search Ads), output MUST comply with these platform limits and structural requirements. Do not output any RSA that violates them.\\n\\n### Hard limits per RSA (enforce before responding)\\n\\n- **Headlines:** exactly **15** per RSA, each **≤ 30 characters** (count characters, including spaces). Render as `1. ... (NN chars)` so the reader can verify.\\n- **Descriptions:** exactly **4** per RSA, each **≤ 90 characters**.\\n- **Paths:** up to 2 path fields, each **≤ 15 characters**.\\n- **Final URL:** present, https.\\n- **Pinning:** state any pinned positions explicitly. Default = unpinned unless user asks.\\n- **Per-account guardrail:** Google enforces **3 RSAs max per ad group**. When the user asks for >3, group them by ad group.\\n\\n### Required sidecar artifacts (always include with RSA request)\\n\\n1. **Ad group structure**, labeled `Ad group structure:` — list each ad group with its theme, target keywords (match types), and which RSAs map to it.\\n2. **Negative keyword list**, labeled `Negative keywords:` — minimum **8** entries, group-level vs campaign-level called out.\\n3. **Sitelinks** (≥ 4), **Callouts** (≥ 4 ≤25 chars), **Structured snippets** if relevant.\\n\\n### Medical / CFM compliance (when product context indicates pt-BR medical practice)\\n\\nIf `.agents/product-marketing.md` indicates a Brazilian medical practice (CFM-regulated), the following terms are **forbidden** in headlines, descriptions, sitelinks, and callouts:\\n\\n- Superlatives: `#1`, `melhor`, `o melhor`, `melhor do brasil`, `top`, `referência`\\n- Outcome promises: `garantido`, `garantia`, `cura`, `cura definitiva`, `100%`, `resultado garantido`, `livre da dor`\\n- Comparative claims vs other doctors/clinics\\n\\nUse neutral framing: `atendimento`, `consulta`, `avaliação`, `segunda opinião`, `agende sua consulta`, `tire suas dúvidas`. Geo modifier (`Porto Alegre`, `POA`, `Zona Sul POA`) required where the prompt specifies a region.\\n\\n### Output ORDER (mandatory — emit in this order to avoid truncation)\\n\\n1. **Ad group structure** (short)\\n2. **Negative keywords** (≥8, MANDATORY — emit BEFORE RSAs so it isn't dropped if output runs long)\\n3. **Sitelinks** (≥4)\\n4. **Callouts** (≥4)\\n5. **RSA1, RSA2, RSA3** (largest section, last — safe to truncate gracefully)\\n\\n### Output template (mandatory shape)\\n\\n```\\nAd group structure:\\n- AG1 [theme]: keywords (match types) → RSA1, RSA2\\n- AG2 [theme]: ...\\n\\nNegative keywords:\\n Campaign-level:\\n - \\n - \\n (≥4 here)\\n Ad-group level:\\n - AG1: , \\n - AG2: , \\n (≥4 more here — TOTAL ≥8 entries)\\n\\nSitelinks (≥4):\\n - | <desc1 (≤35)> | <desc2 (≤35)> | URL\\n\\nCallouts (≥4, each ≤25 chars):\\n - <callout>\\n\\nRSA1 — [ad group name]\\n Final URL: https://...\\n Path1: ... Path2: ...\\n Headlines (15, each ≤30 chars):\\n 1. <headline> (NN chars)\\n ...\\n 15. <headline> (NN chars)\\n Descriptions (4, each ≤90 chars):\\n 1. <description> (NN chars)\\n ...\\n 4. <description> (NN chars)\\n Pinning: H1=none; H2=none; ... (or explicit pins)\\n\\nRSA2 — ...\\nRSA3 — ...\\n```\\n\\n### Self-check before responding\\n\\nBefore sending the output, run this checklist mentally:\\n\\n- [ ] Each RSA has exactly 15 headlines, exactly 4 descriptions.\\n- [ ] Every headline is ≤30 chars; every description is ≤90 chars. Character counts printed.\\n- [ ] Negative keyword list labeled and ≥8 entries.\\n- [ ] Ad group structure labeled.\\n- [ ] If medical (CFM): no forbidden superlative/outcome words; geo modifier present where required; language is pt-BR.\\n\\nIf any check fails, rewrite before responding. Do not ship partial RSAs.\\n\\n---\\n\\n## Common Mistakes to Avoid\\n\\n### Strategy\\n- Launching without conversion tracking\\n- Too many campaigns (fragmenting budget)\\n- Not giving algorithms enough learning time\\n- Optimizing for wrong metric\\n\\n### Targeting\\n- Audiences too narrow or too broad\\n- Not excluding existing customers\\n- Overlapping audiences competing\\n\\n### Creative\\n- Only one ad per ad set\\n- Not refreshing creative (fatigue)\\n- Mismatch between ad and landing page\\n\\n### Budget\\n- Spreading too thin across campaigns\\n- Making big budget changes (disrupts learning)\\n- Stopping campaigns during learning phase\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What platform(s) are you currently running or want to start with?\\n2. What's your monthly ad budget?\\n3. What does a successful conversion look like (and what's it worth)?\\n4. Do you have existing creative assets or need to create them?\\n5. What landing page will ads point to?\\n6. Do you have pixel/conversion tracking set up?\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md). Key advertising platforms:\\n\\n| Platform | Best For | MCP | Guide |\\n|----------|----------|:---:|-------|\\n| **Google Ads** | Search intent, high-intent traffic | ✓ | [google-ads.md](../../tools/integrations/google-ads.md) |\\n| **Meta Ads** | Demand gen, visual products, B2C | - | [meta-ads.md](../../tools/integrations/meta-ads.md) |\\n| **LinkedIn Ads** | B2B, job title targeting | - | [linkedin-ads.md](../../tools/integrations/linkedin-ads.md) |\\n| **TikTok Ads** | Younger demographics, video | - | [tiktok-ads.md](../../tools/integrations/tiktok-ads.md) |\\n\\nFor tracking setup, see [references/conversion-tracking.md](references/conversion-tracking.md), [ga4.md](../../tools/integrations/ga4.md), [segment.md](../../tools/integrations/segment.md)\\n\\n---\\n\\n## Related Skills\\n\\n- **ad-creative**: For generating and iterating ad headlines, descriptions, and creative at scale\\n- **copywriting**: For landing page copy that converts ad traffic\\n- **analytics**: For proper conversion tracking setup\\n- **ab-testing**: For landing page testing to improve ROAS\\n- **cro**: For optimizing post-click conversion rates\"},\n {\"name\":\"ai-seo\",\"category\":\"SEO\",\"description\":\"\\\"When the user wants to optimize content for AI search engines, get cited by LLMs, or appear in AI-generated answers. Also use when the user mentions 'AI SEO,' 'AEO,' 'GEO,' 'LLMO,' 'answer engine optimization,' 'generative engine optimization,' 'LLM optimization,' 'AI Overviews,' 'optimize for ChatGPT,' 'optimize for Perplexity,' 'AI citations,' 'AI visibility,' 'zero-click search,' 'how do I show up in AI answers,' 'LLM mentions,' or 'optimize for Claude/Gemini.' Use this whenever someone wants their content to be cited or surfaced by AI assistants and AI search engines. For traditional technical and on-page SEO audits, see seo-audit. For structured data implementation, see schema.\\\"\",\"version\":\"2.0.1\",\"body\":\"# AI SEO\\n\\nYou are an expert in AI search optimization — the practice of making content discoverable, extractable, and citable by AI systems including Google AI Overviews, ChatGPT, Perplexity, Claude, Gemini, and Copilot. Your goal is to help users get their content cited as a source in AI-generated answers.\\n\\n## Before Starting\\n\\n**Check for product marketing context first:**\\nIf `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.\\n\\nGather this context (ask if not provided):\\n\\n### 1. Current AI Visibility\\n- Do you know if your brand appears in AI-generated answers today?\\n- Have you checked ChatGPT, Perplexity, or Google AI Overviews for your key queries?\\n- What queries matter most to your business?\\n\\n### 2. Content & Domain\\n- What type of content do you produce? (Blog, docs, comparisons, product pages)\\n- What's your domain authority / traditional SEO strength?\\n- Do you have existing structured data (schema markup)?\\n\\n### 3. Goals\\n- Get cited as a source in AI answers?\\n- Appear in Google AI Overviews for specific queries?\\n- Compete with specific brands already getting cited?\\n- Optimize existing content or create new AI-optimized content?\\n\\n### 4. Competitive Landscape\\n- Who are your top competitors in AI search results?\\n- Are they being cited where you're not?\\n\\n---\\n\\n## How AI Search Works\\n\\n### The AI Search Landscape\\n\\n| Platform | How It Works | Source Selection |\\n|----------|-------------|----------------|\\n| **Google AI Overviews** | Summarizes top-ranking pages | Strong correlation with traditional rankings |\\n| **ChatGPT (with search)** | Searches web, cites sources | Draws from wider range, not just top-ranked |\\n| **Perplexity** | Always cites sources with links | Favors authoritative, recent, well-structured content |\\n| **Gemini** | Google's AI assistant | Pulls from Google index + Knowledge Graph |\\n| **Copilot** | Bing-powered AI search | Bing index + authoritative sources |\\n| **Claude** | Brave Search (when enabled) | Training data + Brave search results |\\n\\nFor a deep dive on how each platform selects sources and what to optimize per platform, see [references/platform-ranking-factors.md](references/platform-ranking-factors.md).\\n\\n### Key Difference from Traditional SEO\\n\\nTraditional SEO gets you ranked. AI SEO gets you **cited**.\\n\\nIn traditional search, you need to rank on page 1. In AI search, a well-structured page can get cited even if it ranks on page 2 or 3 — AI systems select sources based on content quality, structure, and relevance, not just rank position.\\n\\n**Critical stats:**\\n- AI Overviews appear in ~45% of Google searches\\n- AI Overviews reduce clicks to websites by up to 58%\\n- Brands are 6.5x more likely to be cited via third-party sources than their own domains\\n- Optimized content gets cited 3x more often than non-optimized\\n- Statistics and citations boost visibility by 40%+ across queries\\n\\n### Google's Official Stance vs. Multi-Platform Reality\\n\\nThis is important to read once before doing anything else.\\n\\n**Google's position** ([AI features optimization guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)):\\n> \\\"The best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems.\\\"\\n\\nGoogle explicitly says:\\n- **No special markup or files are required** for AI Overviews or AI Mode\\n- **Don't chunk content for AI** — write for people, organize with normal headings and paragraphs\\n- **Don't write separate content for AI** — that risks \\\"scaled content abuse\\\" spam policy\\n- **Helpful, reliable, people-first content** wins — same E-E-A-T standards as regular Search\\n- **No AI-specific Search Console reporting** — use standard SEO metrics\\n\\n**Other AI engines (ChatGPT, Claude, Perplexity, Copilot) behave differently:**\\n- They actively reward extractable structure — passages, FAQs, comparison tables, definition blocks\\n- They parse `llms.txt`, structured pricing pages, and machine-readable files when present\\n- They cite third-party sources (Reddit, Wikipedia, review sites) more heavily than top-ranked pages\\n\\n**What this means for the work:**\\n- The structural patterns in this skill (40–60 word answer blocks, FAQ schema, comparison tables) help **non-Google AI engines** materially. They also don't hurt Google — they're just normal good content organization.\\n- For Google AI Overviews / AI Mode specifically: optimize for people and core Search, full stop. Strong E-E-A-T, original information, semantic HTML, clean indexability.\\n- For ChatGPT/Claude/Perplexity: layer on the extractable structure + llms.txt + machine-readable files.\\n\\nWhen in doubt, default to \\\"write for people, organize for clarity\\\" — that satisfies both camps.\\n\\n### Query Fan-Out (Google AI Search)\\n\\nGoogle's AI features don't just answer the one query a user typed — they generate **concurrent, related queries** under the hood and retrieve results for each.\\n\\nGoogle's own example: a user asking \\\"how to fix lawns\\\" triggers fan-out queries about herbicides, chemical-free removal, weed prevention, etc. The AI synthesizes across all of them.\\n\\n**Implications:**\\n- Single-page-per-keyword targeting is less effective. Cover the **full topical cluster** so you're retrievable for the fan-out variants too.\\n- Long-tail intent matters less than topical authority — Google's AI systems understand synonyms and semantic equivalence.\\n- A page that comprehensively answers a parent topic (with sub-questions covered) will be retrieved more often than narrow per-query pages.\\n\\n**Action**: when planning content, brainstorm the 5–10 related queries the AI is likely to fan out to and make sure your content (or your site as a whole) covers them.\\n\\n---\\n\\n## AI Visibility Audit\\n\\nBefore optimizing, assess your current AI search presence.\\n\\n### Step 1: Check AI Answers for Your Key Queries\\n\\nTest 10-20 of your most important queries across platforms:\\n\\n| Query | Google AI Overview | ChatGPT | Perplexity | You Cited? | Competitors Cited? |\\n|-------|:-----------------:|:-------:|:----------:|:----------:|:-----------------:|\\n| [query 1] | Yes/No | Yes/No | Yes/No | Yes/No | [who] |\\n| [query 2] | Yes/No | Yes/No | Yes/No | Yes/No | [who] |\\n\\n**Query types to test:**\\n- \\\"What is [your product category]?\\\"\\n- \\\"Best [product category] for [use case]\\\"\\n- \\\"[Your brand] vs [competitor]\\\"\\n- \\\"How to [problem your product solves]\\\"\\n- \\\"[Your product category] pricing\\\"\\n\\n### Step 2: Analyze Citation Patterns\\n\\nWhen your competitors get cited and you don't, examine:\\n- **Content structure** — Is their content more extractable?\\n- **Authority signals** — Do they have more citations, stats, expert quotes?\\n- **Freshness** — Is their content more recently updated?\\n- **Schema markup** — Do they have structured data you're missing?\\n- **Third-party presence** — Are they cited via Wikipedia, Reddit, review sites?\\n\\n### Step 3: Content Extractability Check\\n\\nFor each priority page, verify:\\n\\n| Check | Pass/Fail |\\n|-------|-----------|\\n| Clear definition in first paragraph? | |\\n| Self-contained answer blocks (work without surrounding context)? | |\\n| Statistics with sources cited? | |\\n| Comparison tables for \\\"[X] vs [Y]\\\" queries? | |\\n| FAQ section with natural-language questions? | |\\n| Schema markup (FAQ, HowTo, Article, Product)? | |\\n| Expert attribution (author name, credentials)? | |\\n| Recently updated (within 6 months)? | |\\n| Heading structure matches query patterns? | |\\n| AI bots allowed in robots.txt? | |\\n\\n### Step 4: AI Bot Access Check\\n\\nVerify your robots.txt allows AI crawlers. Each AI platform has its own bot, and blocking it means that platform can't cite you:\\n\\n- **GPTBot** and **ChatGPT-User** — OpenAI (ChatGPT)\\n- **PerplexityBot** — Perplexity\\n- **ClaudeBot** and **anthropic-ai** — Anthropic (Claude)\\n- **Google-Extended** — Google Gemini and AI Overviews\\n- **Bingbot** — Microsoft Copilot (via Bing)\\n\\nCheck your robots.txt for `Disallow` rules targeting any of these. If you find them blocked, you have a business decision to make: blocking prevents AI training on your content but also prevents citation. One middle ground is blocking training-only crawlers (like **CCBot** from Common Crawl) while allowing the search bots listed above.\\n\\nSee [references/platform-ranking-factors.md](references/platform-ranking-factors.md) for the full robots.txt configuration.\\n\\n---\\n\\n## Optimization Strategy\\n\\n### The Three Pillars\\n\\n```\\n1. Structure (make it extractable)\\n2. Authority (make it citable)\\n3. Presence (be where AI looks)\\n```\\n\\n### Pillar 1: Structure — Make Content Extractable\\n\\nAI systems extract passages, not pages. Every key claim should work as a standalone statement.\\n\\n**Content block patterns:**\\n- **Definition blocks** for \\\"What is X?\\\" queries\\n- **Step-by-step blocks** for \\\"How to X\\\" queries\\n- **Comparison tables** for \\\"X vs Y\\\" queries\\n- **Pros/cons blocks** for evaluation queries\\n- **FAQ blocks** for common questions\\n- **Statistic blocks** with cited sources\\n\\nFor detailed templates for each block type, see [references/content-patterns.md](references/content-patterns.md).\\n\\n**Structural rules:**\\n- Lead every section with a direct answer (don't bury it)\\n- Keep key answer passages to 40-60 words (optimal for snippet extraction)\\n- Use H2/H3 headings that match how people phrase queries\\n- Tables beat prose for comparison content\\n- Numbered lists beat paragraphs for process content\\n- Each paragraph should convey one clear idea\\n\\n### Pillar 2: Authority — Make Content Citable\\n\\nAI systems prefer sources they can trust. Build citation-worthiness.\\n\\n**The Princeton GEO research** (KDD 2024, studied across Perplexity.ai) ranked 9 optimization methods:\\n\\n| Method | Visibility Boost | How to Apply |\\n|--------|:---------------:|--------------|\\n| **Cite sources** | +40% | Add authoritative references with links |\\n| **Add statistics** | +37% | Include specific numbers with sources |\\n| **Add quotations** | +30% | Expert quotes with name and title |\\n| **Authoritative tone** | +25% | Write with demonstrated expertise |\\n| **Improve clarity** | +20% | Simplify complex concepts |\\n| **Technical terms** | +18% | Use domain-specific terminology |\\n| **Unique vocabulary** | +15% | Increase word diversity |\\n| **Fluency optimization** | +15-30% | Improve readability and flow |\\n| ~~Keyword stuffing~~ | **-10%** | **Actively hurts AI visibility** |\\n\\n**Best combination:** Fluency + Statistics = maximum boost. Low-ranking sites benefit even more — up to 115% visibility increase with citations.\\n\\n**Statistics and data** (+37-40% citation boost)\\n- Include specific numbers with sources\\n- Cite original research, not summaries of research\\n- Add dates to all statistics\\n- Original data beats aggregated data\\n\\n**Expert attribution** (+25-30% citation boost)\\n- Named authors with credentials\\n- Expert quotes with titles and organizations\\n- \\\"According to [Source]\\\" framing for claims\\n- Author bios with relevant expertise\\n\\n**Freshness signals**\\n- \\\"Last updated: [date]\\\" prominently displayed\\n- Regular content refreshes (quarterly minimum for competitive topics)\\n- Current year references and recent statistics\\n- Remove or update outdated information\\n\\n**E-E-A-T alignment**\\n- First-hand experience demonstrated\\n- Specific, detailed information (not generic)\\n- Transparent sourcing and methodology\\n- Clear author expertise for the topic\\n\\n### Pillar 3: Presence — Be Where AI Looks\\n\\nAI systems don't just cite your website — they cite where you appear.\\n\\n**Third-party sources matter more than your own site:**\\n- Wikipedia mentions (7.8% of all ChatGPT citations)\\n- Reddit discussions (1.8% of ChatGPT citations)\\n- Industry publications and guest posts\\n- Review sites (G2, Capterra, TrustRadius for B2B SaaS)\\n- YouTube (frequently cited by Google AI Overviews)\\n- Quora answers\\n\\n**Actions:**\\n- Ensure your Wikipedia page is accurate and current\\n- Participate authentically in Reddit communities\\n- Get featured in industry roundups and comparison articles\\n- Maintain updated profiles on relevant review platforms\\n- Create YouTube content for key how-to queries\\n- Answer relevant Quora questions with depth\\n\\n### Machine-Readable Files for AI Agents\\n\\n> **Google's stance**: not required for AI Overviews or AI Mode. Their guide explicitly says you don't need new markup, AI files, or markdown to appear in generative AI search.\\n>\\n> **Why include them anyway**: non-Google AI engines (ChatGPT, Claude, Perplexity) and autonomous buying agents do reward extractable structure. The files below help with those engines without harming Google.\\n\\nAI agents aren't just answering questions — they're becoming buyers. When an AI agent evaluates tools on behalf of a user, it needs structured, parseable information. If your pricing is locked in a JavaScript-rendered page or a \\\"contact sales\\\" wall, agents will skip you and recommend competitors whose information they can actually read.\\n\\nAdd these machine-readable files to your site root:\\n\\n**`/pricing.md` or `/pricing.txt`** — Structured pricing data for AI agents\\n\\n```markdown\\n# Pricing — [Your Product Name]\\n\\n## Free\\n- Price: $0/month\\n- Limits: 100 emails/month, 1 user\\n- Features: Basic templates, API access\\n\\n## Pro\\n- Price: $29/month (billed annually) | $35/month (billed monthly)\\n- Limits: 10,000 emails/month, 5 users\\n- Features: Custom domains, analytics, priority support\\n\\n## Enterprise\\n- Price: Custom — contact sales@example.com\\n- Limits: Unlimited emails, unlimited users\\n- Features: SSO, SLA, dedicated account manager\\n```\\n\\n**Why this matters now:**\\n- AI agents increasingly compare products programmatically before a human ever visits your site\\n- Opaque pricing gets filtered out of AI-mediated buying journeys\\n- A simple markdown file is trivially parseable by any LLM — no rendering, no JavaScript, no login walls\\n- Same principle as `robots.txt` (for crawlers), `llms.txt` (for AI context), and `AGENTS.md` (for agent capabilities)\\n\\n**Best practices:**\\n- Use consistent units (monthly vs. annual, per-seat vs. flat)\\n- Include specific limits and thresholds, not just feature names\\n- List what's included at each tier, not just what's different\\n- Keep it updated — stale pricing is worse than no file\\n- Link to it from your sitemap and main pricing page\\n\\n**`/llms.txt`** — Context file for AI systems (see [llmstxt.org](https://llmstxt.org))\\n\\nIf you don't have one yet, add an `llms.txt` that gives AI systems a quick overview of what your product does, who it's for, and links to key pages (including your pricing).\\n\\n### Schema Markup for AI\\n\\nStructured data helps AI systems understand your content. Key schemas:\\n\\n| Content Type | Schema | Why It Helps |\\n|-------------|--------|-------------|\\n| Articles/Blog posts | `Article`, `BlogPosting` | Author, date, topic identification |\\n| How-to content | `HowTo` | Step extraction for process queries |\\n| FAQs | `FAQPage` | Direct Q&A extraction |\\n| Products | `Product` | Pricing, features, reviews |\\n| Comparisons | `ItemList` | Structured comparison data |\\n| Reviews | `Review`, `AggregateRating` | Trust signals |\\n| Organization | `Organization` | Entity recognition |\\n\\nContent with proper schema shows 30-40% higher AI visibility on non-Google AI engines. **Google's note**: structured data is \\\"not required for generative AI search\\\" but is recommended for overall SEO strategy. For implementation, use the **schema** skill.\\n\\n---\\n\\n## Agentic Experiences\\n\\nBeyond AI search engines summarizing content, autonomous agents are starting to access sites directly — clicking, reading, comparing, even buying on behalf of users. Google's guide flags this as an emerging category to plan for.\\n\\n**How agents access your site:**\\n- **Visual rendering** — they screenshot/read the page like a user would\\n- **DOM inspection** — they parse the page's HTML structure\\n- **Accessibility tree** — they rely on the same semantic information assistive tech uses (labels, roles, landmarks, headings)\\n\\n**What to do:**\\n- **Render meaningful content without heavy JS gymnastics** — if the page is blank until 4 frameworks finish loading, agents see blank\\n- **Semantic HTML** — use `<main>`, `<nav>`, `<article>`, `<button>`, proper heading hierarchy, `alt` text on images\\n- **Clean accessibility tree** — every interactive element labelled; ARIA used correctly (or not at all when native HTML suffices)\\n- **Stable selectors / predictable layouts** — agents struggle with sites that re-render every interaction\\n- **Visible pricing, specs, contact info** — anything an agent would need to make a buying recommendation should be on a public, indexable page (this is where `/pricing.md` and similar files help)\\n\\n**Emerging — Universal Commerce Protocol (UCP):**\\nGoogle references UCP as a forthcoming protocol that will give agents standardized hooks for commerce interactions (catalog discovery, pricing, checkout). Watch for adoption; for now, the structural recommendations above are the precursor.\\n\\nFor ecom and local business specifically, Google highlights:\\n- **Merchant Center feeds** + **Google Business Profile** for product/service visibility in AI Search\\n- **Business Agent** for conversational customer engagement (where applicable)\\n\\n---\\n\\n## Content Types That Get Cited Most\\n\\nNot all content is equally citable. Prioritize these formats:\\n\\n| Content Type | Citation Share | Why AI Cites It |\\n|-------------|:------------:|----------------|\\n| **Comparison articles** | ~33% | Structured, balanced, high-intent |\\n| **Definitive guides** | ~15% | Comprehensive, authoritative |\\n| **Original research/data** | ~12% | Unique, citable statistics |\\n| **Best-of/listicles** | ~10% | Clear structure, entity-rich |\\n| **Product pages** | ~10% | Specific details AI can extract |\\n| **How-to guides** | ~8% | Step-by-step structure |\\n| **Opinion/analysis** | ~10% | Expert perspective, quotable |\\n\\n**Underperformers for AI citation:**\\n- Generic blog posts without structure\\n- Thin product pages with marketing fluff\\n- Gated content (AI can't access it)\\n- Content without dates or author attribution\\n- PDF-only content (harder for AI to parse)\\n\\n---\\n\\n## Monitoring AI Visibility\\n\\n### What to Track\\n\\n| Metric | What It Measures | How to Check |\\n|--------|-----------------|-------------|\\n| AI Overview presence | Do AI Overviews appear for your queries? | Manual check or Semrush/Ahrefs |\\n| Brand citation rate | How often you're cited in AI answers | AI visibility tools (see below) |\\n| Share of AI voice | Your citations vs. competitors | Peec AI, Otterly, ZipTie |\\n| Citation sentiment | How AI describes your brand | Manual review + monitoring tools |\\n| Source attribution | Which of your pages get cited | Track referral traffic from AI sources |\\n\\n### AI Visibility Monitoring Tools\\n\\n| Tool | Coverage | Best For |\\n|------|----------|----------|\\n| **Otterly AI** | ChatGPT, Perplexity, Google AI Overviews | Share of AI voice tracking |\\n| **Peec AI** | ChatGPT, Gemini, Perplexity, Claude, Copilot+ | Multi-platform monitoring at scale |\\n| **ZipTie** | Google AI Overviews, ChatGPT, Perplexity | Brand mention + sentiment tracking |\\n| **LLMrefs** | ChatGPT, Perplexity, AI Overviews, Gemini | SEO keyword → AI visibility mapping |\\n\\n### DIY Monitoring (No Tools)\\n\\nMonthly manual check:\\n1. Pick your top 20 queries\\n2. Run each through ChatGPT, Perplexity, and Google\\n3. Record: Are you cited? Who is? What page?\\n4. Log in a spreadsheet, track month-over-month\\n\\n### Search Console expectations\\n\\nGoogle's guide is explicit: **there is no AI-specific Search Console reporting**. AI Overviews and AI Mode use core Search ranking, so the standard Search Console reports (Performance, Coverage, Core Web Vitals) are still what you measure with for Google. The third-party tools above are the only way to see cross-platform AI citation behavior.\\n\\n---\\n\\n## What NOT to Do\\n\\nGoogle's guide calls these out explicitly — they hurt across both traditional Search and AI features.\\n\\n1. **Write separate content \\\"for AI\\\"**. Same content should serve people and AI. Writing variants targeted at AI systems risks the **scaled content abuse spam policy** — Google's words.\\n2. **Chunk pages into AI-bait fragments**. Google's guide is direct: *\\\"Don't break your content into tiny pieces for AI to better understand it.\\\"* Use normal paragraph + heading structure.\\n3. **Generate at scale for ranking manipulation**. AI-generated content is fine *if* it meets Search Essentials and spam policies. Mass-producing thin variations does not.\\n4. **Pursue inauthentic mentions**. Don't fabricate citations or bulk-spam Reddit/Wikipedia for AI visibility. Real participation only.\\n5. **Block AI crawlers if you want citation**. Blocking GPTBot, PerplexityBot, ClaudeBot, Google-Extended means those engines literally cannot cite you. Block training-only crawlers (CCBot) if you must, not the search-and-cite ones.\\n6. **Hide your main content behind JS that doesn't render**. Both core Search and AI agents need to see your content; JS-only rendering loses both audiences.\\n7. **Skip E-E-A-T fundamentals**. Author identity, first-hand experience, expertise signals, transparent sourcing — Google's guide leans heavily on these for AI features.\\n\\n---\\n\\n## AI SEO by Content Type\\n\\nFor tactical guidance on SaaS product pages, blog content, comparison/alternative pages, documentation, and local/ecom (Google's emphasis on Merchant Center + Business Profile), see [references/content-types.md](references/content-types.md).\\n\\n---\\n\\n## Common Mistakes\\n\\n- **Ignoring AI search entirely** — ~45% of Google searches now show AI Overviews, and ChatGPT/Perplexity are growing fast\\n- **Treating AI SEO as separate from SEO** — Good traditional SEO is the foundation; AI SEO adds structure and authority on top\\n- **Writing for AI, not humans** — If content reads like it was written to game an algorithm, it won't get cited or convert\\n- **No freshness signals** — Undated content loses to dated content because AI systems weight recency heavily. Show when content was last updated\\n- **Gating all content** — AI can't access gated content. Keep your most authoritative content open\\n- **Ignoring third-party presence** — You may get more AI citations from a Wikipedia mention than from your own blog\\n- **No structured data** — Schema markup gives AI systems structured context about your content\\n- **Keyword stuffing** — Unlike traditional SEO where it's just ineffective, keyword stuffing actively reduces AI visibility by 10% (Princeton GEO study)\\n- **Hiding pricing behind \\\"contact sales\\\" or JS-rendered pages** — AI agents evaluating your product on behalf of buyers can't parse what they can't read. Add a `/pricing.md` file\\n- **Blocking AI bots** — If GPTBot, PerplexityBot, or ClaudeBot are blocked in robots.txt, those platforms can't cite you\\n- **Generic content without data** — \\\"We're the best\\\" won't get cited. \\\"Our customers see 3x improvement in [metric]\\\" will\\n- **Forgetting to monitor** — You can't improve what you don't measure. Check AI visibility monthly at minimum\\n\\n---\\n\\n## Tool Integrations\\n\\nFor implementation, see the [tools registry](../../tools/REGISTRY.md).\\n\\n| Tool | Use For |\\n|------|---------|\\n| `semrush` | AI Overview tracking, keyword research, content gap analysis |\\n| `ahrefs` | Backlink analysis, content explorer, AI Overview data |\\n| `gsc` | Search Console performance data, query tracking |\\n| `ga4` | Referral traffic from AI sources |\\n\\n---\\n\\n## Task-Specific Questions\\n\\n1. What are your top 10-20 most important queries?\\n2. Have you checked if AI answers exist for those queries today?\\n3. Do you have structured data (schema markup) on your site?\\n4. What content types do you publish? (Blog, docs, comparisons, etc.)\\n5. Are competitors being cited by AI where you're not?\\n6. Do you have a Wikipedia page or presence on review sites?\\n\\n---\\n\\n## Related Skills\\n\\n- **seo-audit**: For traditional technical and on-page SEO audits\\n- **schema**: For implementing structured data that helps AI understand your content\\n- **content-strategy**: For planning what content to create\\n- **competitors**: For building comparison pages that get cited\\n- **programmatic-seo**: For building SEO pages at scale\\n- **copywriting**: For writing content that's both human-readable and AI-extractable\"}\n];\n",[],[352,552,554],"eb52f4d2-5d6d-11f1-95b0-1607ee4eb77e",["D",1780286210534],[444,445],[352,552],["D",1780286210534],[443],[352],"d86ff084-5b4e-11f1-80fc-ee650bb23af1","youtube.ts","d86ff138-5b4e-11f1-80fc-ee650bb23af1",536,["D",1780052962192],"// YouTubeRenderer — per-integration renderer for the 'youtube' integration.\nexport const RENDERER = `\nexport function YouTubeRenderer({data, endpoint}) {\n const YT = '#ff0033';\n const errMsg = data?.error && (typeof data.error === 'string' ? data.error : (data.error.message || JSON.stringify(data.error)));\n if (errMsg) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-start gap-3 border border-error/30'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-on-surface mb-1'}, 'YouTube API error'),\n h('div', {className:'text-[12px] text-text-muted'}, errMsg)\n )\n );\n\n const items = Array.isArray(data?.items) ? data.items : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No results.');\n\n const fmtCount = (v) => {\n const n = parseInt(v, 10);\n if (isNaN(n)) return '-';\n if (n < 1000) return String(n);\n if (n < 1000000) return (n/1000).toFixed(n < 10000 ? 1 : 0) + 'K';\n if (n < 1000000000) return (n/1000000).toFixed(n < 10000000 ? 1 : 0) + 'M';\n return (n/1000000000).toFixed(1) + 'B';\n };\n const fmtDur = (iso) => {\n if (!iso || typeof iso !== 'string') return '';\n const m = iso.match(/PT(?:(\\\\d+)H)?(?:(\\\\d+)M)?(?:(\\\\d+)S)?/);\n if (!m) return '';\n const hh = parseInt(m[1]||'0',10), mm = parseInt(m[2]||'0',10), ss = parseInt(m[3]||'0',10);\n const p2 = (x) => (x < 10 ? '0'+x : ''+x);\n return hh ? (hh + ':' + p2(mm) + ':' + p2(ss)) : (mm + ':' + p2(ss));\n };\n const thumb = (sn) => sn?.thumbnails?.medium?.url || sn?.thumbnails?.high?.url || sn?.thumbnails?.default?.url || '';\n const vidId = (it) => (it.id && typeof it.id === 'object' ? (it.id.videoId || '') : '');\n const chId = (it) => (typeof it.id === 'object' ? (it.id.channelId || '') : (typeof it.id === 'string' ? it.id : ''));\n const chip = (icon, val, color) => h('span', {className:'inline-flex items-center gap-1 text-[10px] font-mono-data text-on-surface-variant'},\n h(Icon, {name:icon, size:11, style:{color: color || YT}}), fmtCount(val)\n );\n\n // --- CHANNELS ----------------------------------------------------------\n if (endpoint === 'channels') {\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(280px, 1fr))'}},\n items.map(ch => {\n const sn = ch.snippet || {}, st = ch.statistics || {};\n const url = sn.customUrl ? 'https://youtube.com/' + sn.customUrl : ('https://youtube.com/channel/' + (typeof ch.id === 'string' ? ch.id : chId(ch)));\n const avatar = thumb(sn);\n return h('a', {key:ch.id?.channelId || ch.id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl p-md flex items-center gap-3 hover:border-primary/40 transition-all group'},\n avatar ? h('img', {src:avatar, className:'w-14 h-14 rounded-full border border-border-subtle flex-shrink-0'})\n : h('div', {className:'w-14 h-14 rounded-full flex items-center justify-center font-bold text-[16px] flex-shrink-0', style:{background:colorFromName(sn.title)+'33', color:colorFromName(sn.title)}}, initialsOf(sn.title)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate flex items-center gap-1'}, sn.title || '(channel)',\n h(Icon, {name:'open_in_new', size:12, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity'})),\n sn.description && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5'}, sn.description),\n h('div', {className:'flex items-center gap-3 mt-1.5'},\n chip('group', st.subscriberCount),\n chip('visibility', st.viewCount, '#94a3b8'),\n st.videoCount != null && chip('movie', st.videoCount, '#94a3b8')\n )\n )\n );\n })\n );\n }\n\n // --- PLAYLISTS ---------------------------------------------------------\n if (endpoint === 'playlists') {\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(300px, 1fr))'}},\n items.map(pl => {\n const sn = pl.snippet || {}, cd = pl.contentDetails || {};\n const id = typeof pl.id === 'string' ? pl.id : '';\n const url = id ? ('https://youtube.com/playlist?list=' + id) : '#';\n const img = thumb(sn);\n return h('a', {key:id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-primary/40 transition-all group'},\n h('div', {className:'relative aspect-video bg-surface-container-high overflow-hidden'},\n img ? h('img', {src:img, className:'w-full h-full object-cover'})\n : h('div', {className:'w-full h-full flex items-center justify-center'}, h(Icon, {name:'playlist_play', size:32, className:'text-text-muted'})),\n cd.itemCount != null && h('div', {className:'absolute right-1.5 bottom-1.5 px-1.5 py-0.5 rounded text-[10px] font-mono-data text-white flex items-center gap-1', style:{background:'rgba(0,0,0,0.78)'}},\n h(Icon, {name:'playlist_play', size:12}), cd.itemCount + ' videos')\n ),\n h('div', {className:'p-md flex flex-col gap-1'},\n h('div', {className:'text-[13px] font-bold text-on-surface leading-snug', style:{display:'-webkit-box', WebkitLineClamp:2, WebkitBoxOrient:'vertical', overflow:'hidden'}}, sn.title || '(untitled playlist)'),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted'},\n h('span', {className:'truncate flex items-center gap-1'}, h(Icon, {name:'account_circle', size:12}), sn.channelTitle || '-'),\n sn.publishedAt && h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(sn.publishedAt))\n )\n )\n );\n })\n );\n }\n\n // --- VIDEOS / SEARCH ---------------------------------------------------\n return h('div', {className:'mt-3 grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(300px, 1fr))'}},\n items.map(it => {\n const sn = it.snippet || {}, st = it.statistics || {}, cd = it.contentDetails || {};\n const id = vidId(it) || (typeof it.id === 'string' ? it.id : '');\n const url = id ? ('https://youtube.com/watch?v=' + id) : '#';\n const img = thumb(sn);\n const dur = fmtDur(cd.duration);\n const hasStats = st.viewCount != null || st.likeCount != null || st.commentCount != null;\n return h('a', {key:id || sn.title, href:url, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-primary/40 transition-all group'},\n h('div', {className:'relative aspect-video bg-surface-container-high overflow-hidden'},\n img ? h('img', {src:img, className:'w-full h-full object-cover'})\n : h('div', {className:'w-full h-full flex items-center justify-center'}, h(Icon, {name:'movie', size:32, className:'text-text-muted'})),\n dur && h('div', {className:'absolute right-1.5 bottom-1.5 px-1.5 py-0.5 rounded text-[10px] font-mono-data text-white', style:{background:'rgba(0,0,0,0.78)'}}, dur),\n h('div', {className:'absolute inset-0 flex items-center justify-center bg-black/0 group-hover:bg-black/40 transition-colors'},\n h('div', {className:'w-12 h-12 rounded-full flex items-center justify-center opacity-0 group-hover:opacity-100 transition-opacity', style:{background:YT}},\n h(Icon, {name:'play_arrow', size:26, style:{color:'#fff'}}))\n )\n ),\n h('div', {className:'p-md flex flex-col gap-1.5 flex-1'},\n h('div', {className:'text-[13px] font-bold text-on-surface leading-snug', style:{display:'-webkit-box', WebkitLineClamp:2, WebkitBoxOrient:'vertical', overflow:'hidden'}}, sn.title || '(untitled)'),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted'},\n h('span', {className:'truncate flex items-center gap-1'}, h(Icon, {name:'account_circle', size:12}), sn.channelTitle || '-'),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(sn.publishedAt))\n ),\n hasStats && h('div', {className:'flex items-center gap-3 pt-1.5 mt-auto border-t border-border-subtle'},\n st.viewCount != null && chip('visibility', st.viewCount, '#94a3b8'),\n st.likeCount != null && chip('thumb_up', st.likeCount, '#4edea3'),\n st.commentCount != null && chip('forum', st.commentCount, '#c0c1ff')\n )\n )\n );\n })\n );\n}\n`;\n",[],[359,568],"d7f7819e-5b4e-11f1-a8fd-ee650bb23af1","wordpress.ts","d7f78266-5b4e-11f1-a8fd-ee650bb23af1",535,["D",1780052961393],"// WordPressRenderer — per-integration renderer for the 'wordpress' integration.\nexport const RENDERER = `\nexport function WordPressRenderer({data, endpoint}) {\n const WP_BLUE = '#21759b';\n const strip = (html) => (html || '').replace(/<[^>]+>/g, '').replace(/&#?\\\\w+;/g, ' ').replace(/\\\\s+/g, ' ').trim();\n const empty = (label) => h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, label);\n const statusMeta = (s) => s === 'publish' ? {label:'Published', color:'#4edea3'}\n : s === 'draft' ? {label:'Draft', color:'#fbbc04'}\n : s === 'pending' ? {label:'Pending', color:'#8083ff'}\n : s === 'future' ? {label:'Scheduled', color:'#7fc2dc'}\n : s === 'private' ? {label:'Private', color:'#d0bcff'}\n : {label: s || 'Unknown', color:'#94a3b8'};\n\n // Error card\n if (data && !Array.isArray(data) && data.error) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-on-surface'}, 'WordPress error'),\n h('div', {className:'text-[12px] text-text-muted mt-0.5'}, strip(String(data.error)))));\n }\n\n // PUBLISH-POST: single created-post success card\n if (endpoint === 'publish-post' && data && !Array.isArray(data) && data.id) {\n const sm = statusMeta(data.status);\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-emerald-400/40'},\n h('div', {className:'flex items-center gap-2 mb-3'},\n h(Icon, {name:'check_circle', size:18, className:'text-emerald-400'}),\n h('div', {className:'text-[13px] font-bold text-on-surface'}, 'Post created'),\n h('span', {className:'ml-auto inline-flex items-center gap-1.5 px-2.5 py-1 rounded-full text-[10px] font-bold uppercase tracking-wide', style:{background:sm.color+'22', color:sm.color, border:'1px solid '+sm.color+'55'}},\n h('span', {className:'w-1.5 h-1.5 rounded-full', style:{background:sm.color}}), sm.label)),\n h('div', {className:'text-[15px] font-bold text-on-surface'}, strip(data.title?.rendered || data.title?.raw) || '(no title)'),\n strip(data.excerpt?.rendered) && h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 line-clamp-2'}, strip(data.excerpt?.rendered).slice(0,220)),\n h('div', {className:'flex items-center gap-4 mt-3 pt-3 border-t border-border-subtle text-[10px] text-text-muted font-mono-data'},\n h('span', {className:'flex items-center gap-1'}, h(Icon,{name:'calendar_today',size:11}), fmtRelative(data.date) || 'now'),\n h('span', {className:'ml-auto'}, '#' + data.id)),\n data.link && h('a', {href:data.link, target:'_blank', rel:'noreferrer', className:'inline-flex items-center gap-1.5 mt-3 text-[12px] font-medium text-secondary hover:underline'},\n h(Icon,{name:'open_in_new',size:13}), 'View post'));\n }\n\n const rows = Array.isArray(data) ? data : [];\n\n // CATEGORIES: chips with counts\n if (endpoint === 'categories') {\n if (!rows.length) return empty('No categories.');\n const sorted = rows.slice().sort((a, b) => (b.count || 0) - (a.count || 0));\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3'}, sorted.length + ' Categories'),\n h('div', {className:'flex flex-wrap gap-2'},\n sorted.map(c => h('a', {key:c.id, href:c.link || '#', target:'_blank', rel:'noreferrer',\n className:'group inline-flex items-center gap-2 px-3 py-1.5 rounded-full bg-surface-container-high border border-border-subtle hover:border-secondary/50 transition-colors'},\n h('span', {className:'text-[12px] font-medium text-on-surface'}, strip(c.name)),\n h('span', {className:'px-1.5 py-0.5 rounded-full text-[10px] font-mono-data font-bold', style:{background:WP_BLUE+'22', color:'#7fc2dc'}}, (c.count || 0).toLocaleString())))));\n }\n\n // MEDIA: thumbnail grid\n if (endpoint === 'media') {\n if (!rows.length) return empty('No media.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(160px, 1fr))'}},\n rows.map(m => {\n const thumb = m.media_details?.sizes?.medium?.source_url || m.media_details?.sizes?.thumbnail?.source_url || m.source_url;\n const isImg = (m.media_type === 'image') || /^image\\\\//.test(m.mime_type || '');\n const size = m.media_details?.filesize;\n return h('a', {key:m.id, href:m.link || m.source_url || '#', target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl overflow-hidden flex flex-col hover:border-secondary/40 transition-all group'},\n h('div', {className:'aspect-square bg-surface-container flex items-center justify-center overflow-hidden'},\n isImg && thumb\n ? h('img', {src:thumb, loading:'lazy', className:'w-full h-full object-cover group-hover:scale-105 transition-transform', alt:strip(m.title?.rendered)})\n : h(Icon, {name: /^video\\\\//.test(m.mime_type||'') ? 'movie' : /^audio\\\\//.test(m.mime_type||'') ? 'audiotrack' : 'insert_drive_file', size:28, className:'text-text-muted'})),\n h('div', {className:'p-2.5'},\n h('div', {className:'text-[11px] font-bold text-on-surface truncate'}, strip(m.title?.rendered) || 'Untitled'),\n h('div', {className:'flex items-center justify-between text-[9px] text-text-muted font-mono-data mt-1'},\n h('span', {className:'uppercase'}, (m.mime_type || '').split('/')[1] || m.media_type || '-'),\n h('span', null, size ? fmtFileSize(size) : fmtRelative(m.date)))));\n })));\n }\n\n // POSTS / PAGES: post cards\n if (rows.length || endpoint === 'posts' || endpoint === 'pages') {\n if (!rows.length) return empty('No ' + (endpoint || 'posts') + '.');\n return h('div', {className:'mt-3 space-y-3'},\n rows.map(p => {\n const title = strip(p.title?.rendered) || '(no title)';\n const sm = statusMeta(p.status);\n const excerpt = strip(p.excerpt?.rendered);\n return h('div', {key:p.id, className:'glass-panel rounded-xl p-md hover:border-secondary/30 transition-all'},\n h('div', {className:'flex items-start justify-between gap-3'},\n h('a', {href:p.link || '#', target:'_blank', rel:'noreferrer', className:'text-[15px] font-bold text-on-surface hover:text-secondary transition-colors flex items-center gap-1.5 group min-w-0'},\n h('span', {className:'truncate'}, title),\n h(Icon, {name:'open_in_new', size:13, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})),\n h('span', {className:'inline-flex items-center gap-1.5 px-2.5 py-1 rounded-full text-[10px] font-bold uppercase tracking-wide flex-shrink-0', style:{background:sm.color+'22', color:sm.color, border:'1px solid '+sm.color+'55'}},\n h('span', {className:'w-1.5 h-1.5 rounded-full', style:{background:sm.color}}), sm.label)),\n excerpt && h('div', {className:'text-[12px] text-on-surface-variant mt-2 line-clamp-2'}, excerpt.slice(0, 220)),\n h('div', {className:'flex items-center gap-4 mt-2.5 pt-2.5 border-t border-border-subtle text-[10px] text-text-muted font-mono-data'},\n h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'calendar_today', size:11}), fmtRelative(p.date)),\n p.author != null && h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'person', size:11}), 'Author #' + p.author),\n h('span', {className:'flex items-center gap-1'}, h(Icon, {name:'comment', size:11}), (p.comment_count || 0).toLocaleString() + (p.comment_count === 1 ? ' comment' : ' comments')),\n h('span', {className:'ml-auto text-on-surface-variant'}, '#' + p.id)));\n }));\n }\n\n // Generic single-object fallback (default/custom endpoints)\n if (data && !Array.isArray(data) && typeof data === 'object') {\n const keys = Object.keys(data).slice(0, 12);\n if (!keys.length) return empty('No data.');\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3'}, endpoint || 'WordPress'),\n h('div', {className:'space-y-1.5'},\n keys.map(k => {\n const v = data[k];\n const val = (v && typeof v === 'object') ? (v.rendered != null ? strip(v.rendered) : JSON.stringify(v).slice(0,120)) : String(v);\n return h('div', {key:k, className:'flex items-start gap-3 text-[12px] border-b border-border-subtle/50 pb-1.5'},\n h('span', {className:'text-text-muted font-mono-data min-w-[120px]'}, k),\n h('span', {className:'text-on-surface break-all'}, val || '-'));\n })));\n }\n return empty('No data.');\n}\n`;\n",[],[359,576],"d793085e-5b4e-11f1-9305-ee650bb23af1","teams.ts","d793091c-5b4e-11f1-9305-ee650bb23af1",534,["D",1780052960743],"// TeamsRenderer — per-integration renderer for the 'teams' integration.\nexport const RENDERER = `\nexport function TeamsRenderer({data, endpoint}) {\n const {useState} = React;\n const TEAMS_PURPLE = '#6264a7';\n const stripHtml = (s) => (s||'').replace(/<[^>]+>/g,'').replace(/ /g,' ').replace(/&/g,'&').replace(/</g,'<').replace(/>/g,'>').trim();\n\n if (data && data.error) {\n const e = data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[14px]'}, h(Icon,{name:'error',size:18}), 'Teams error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 font-mono-data break-words'}, (e.code ? e.code + ': ' : '') + (e.message || String(e))));\n }\n\n // --- WRITE CONFIRMATIONS ---\n if (endpoint === 'send-channel' || endpoint === 'send-chat') {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-3 border border-secondary/30'},\n h('div', {className:'w-10 h-10 rounded-full flex items-center justify-center flex-shrink-0', style:{background:TEAMS_PURPLE+'22',color:TEAMS_PURPLE}}, h(Icon,{name:'check_circle',size:22})),\n h('div', null,\n h('div', {className:'text-[14px] font-bold text-on-surface'}, 'Message sent'),\n h('div', {className:'text-[11px] text-text-muted mt-0.5'}, data && data.createdDateTime ? fmtRelative(data.createdDateTime) : 'Delivered to Microsoft Teams')));\n }\n\n // --- TEAMS (cards) ---\n if (endpoint === 'teams' || !endpoint) {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No teams joined.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n items.map((t,i) => {\n const name = t.displayName || '(untitled team)';\n return h('div', {key:t.id||i, className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/30 transition-all'},\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center font-bold text-[13px] flex-shrink-0', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, name),\n t.description && h('div', {className:'text-[11px] text-text-muted mt-1 line-clamp-2'}, t.description),\n t.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1.5'}, t.id)));\n })));\n }\n\n // --- CHANNELS (list) ---\n if (endpoint === 'channels') {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No channels.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'divide-y divide-border-subtle/50'},\n items.map((c,i) => h('div', {key:c.id||i, className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors'},\n h('div', {className:'w-8 h-8 rounded-lg flex items-center justify-center flex-shrink-0 mt-0.5', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, h(Icon,{name:'tag',size:16})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, c.displayName || '(unnamed)'),\n c.description && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5'}, c.description),\n c.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1'}, c.id))))));\n }\n\n // --- CHATS (list) ---\n if (endpoint === 'chats') {\n const items = Array.isArray(data && data.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No chats.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'divide-y divide-border-subtle/50'},\n items.map((c,i) => {\n const title = c.topic || (c.chatType === 'oneOnOne' ? 'Direct chat' : c.chatType === 'group' ? 'Group chat' : 'Chat');\n return h('div', {key:c.id||i, className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors'},\n h('div', {className:'w-8 h-8 rounded-full flex items-center justify-center flex-shrink-0 mt-0.5', style:{background:TEAMS_PURPLE+'22',color:'#c0c1ff'}}, h(Icon,{name:c.chatType==='group'?'groups':'chat',size:16})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, title),\n h('div', {className:'flex items-center gap-2 mt-0.5'},\n c.chatType && h('span', {className:'text-[10px] text-text-muted'}, c.chatType),\n c.lastUpdatedDateTime && h('span', {className:'text-[10px] text-text-muted font-mono-data'}, fmtRelative(c.lastUpdatedDateTime))),\n c.id && h('div', {className:'text-[9px] text-text-muted/70 font-mono-data truncate mt-1'}, c.id)));\n })));\n }\n\n // --- MESSAGES + CHAT-MESSAGES (thread + inline compose) ---\n const isChat = endpoint === 'chat-messages';\n const msgs = Array.isArray(data && data.value) ? data.value : [];\n const ctx = msgs[0] || {};\n const ctxTeamId = (ctx.channelIdentity && ctx.channelIdentity.teamId) || '';\n const ctxChannelId = (ctx.channelIdentity && ctx.channelIdentity.channelId) || '';\n const ctxChatId = ctx.chatId || '';\n\n const [teamId, setTeamId] = useState(ctxTeamId);\n const [channelId, setChannelId] = useState(ctxChannelId);\n const [chatId, setChatId] = useState(ctxChatId);\n const [draft, setDraft] = useState('');\n const [sending, setSending] = useState(false);\n const [status, setStatus] = useState('');\n\n const needIds = isChat ? !chatId : (!teamId || !channelId);\n const send = async () => {\n const content = draft.trim();\n if (!content || needIds) return;\n setSending(true); setStatus('');\n try {\n const ep = isChat ? 'send-chat' : 'send-channel';\n const body = isChat ? {chatId, content} : {teamId, channelId, content};\n const r = await fetch('/api/integrations/teams/data?endpoint=' + ep, {\n method:'POST', headers:{'Content-Type':'application/json'}, body: JSON.stringify(body) });\n const j = await r.json().catch(() => ({}));\n if (r.ok && !(j && j.error)) { setStatus('sent'); setDraft(''); }\n else setStatus('Error: ' + ((j && j.error && (j.error.message || j.error)) || ('HTTP ' + r.status)));\n } catch (e) { setStatus('Connection error'); }\n setSending(false);\n };\n\n const idInput = (val, set, ph) => h('input', {value:val, onChange:e=>set(e.target.value), placeholder:ph,\n className:'flex-1 min-w-0 bg-surface-container-lowest border border-border-subtle rounded-lg px-2.5 py-1.5 text-[11px] font-mono-data text-on-surface outline-none focus:border-primary'});\n\n const composer = h('div', {className:'p-3.5 border-t border-border-subtle bg-surface-container-lowest/50 space-y-2'},\n needIds && h('div', {className:'flex gap-2'},\n isChat ? idInput(chatId, setChatId, 'Chat ID')\n : [idInput(teamId, setTeamId, 'Team ID'), idInput(channelId, setChannelId, 'Channel ID')]),\n h('div', {className:'flex items-end gap-2'},\n h('textarea', {value:draft, onChange:e=>setDraft(e.target.value), rows:2,\n placeholder: isChat ? 'Reply to this chat...' : 'Reply to this channel...',\n className:'flex-1 bg-surface-container-lowest border border-border-subtle rounded-lg px-3 py-2 text-[13px] text-on-surface outline-none focus:border-primary resize-none'}),\n h('button', {onClick:send, disabled: sending || !draft.trim() || needIds,\n className:'px-3.5 py-2 rounded-lg text-white text-[13px] font-semibold disabled:opacity-50 flex items-center gap-1.5 flex-shrink-0',\n style:{background:TEAMS_PURPLE}},\n h(Icon,{name:'send',size:15}), sending ? 'Sending...' : 'Send')),\n status && h('div', {className:'text-[11px] ' + (status==='sent' ? 'text-secondary' : 'text-error')},\n status==='sent' ? 'Message sent to Teams.' : status));\n\n if (!msgs.length) {\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'text-center text-text-muted text-[13px] py-8'}, 'No messages yet.'),\n composer);\n }\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex items-center gap-2 bg-surface-container-lowest/50'},\n h(Icon,{name:'forum',size:18,style:{color:'#c0c1ff'}}),\n h('h3', {className:'text-[14px] font-bold'}, msgs.length + ' message' + (msgs.length===1?'':'s'))),\n h('div', {className:'divide-y divide-border-subtle/40 max-h-[480px] overflow-y-auto'},\n msgs.map((m,i) => {\n const sender = (m.from && m.from.user && m.from.user.displayName) || (m.from && m.from.application && m.from.application.displayName) || 'System';\n const text = stripHtml(m.body && m.body.content);\n const isSystem = m.messageType && m.messageType !== 'message';\n const col = colorFromName(sender);\n if (isSystem && !text) return null;\n return h('div', {key:m.id||i, className:'flex items-start gap-3 p-3.5'},\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(sender)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className:'text-[13px] font-bold text-on-surface truncate'}, sender),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.createdDateTime))),\n text ? h('div', {className:'text-[13px] text-on-surface-variant mt-0.5 whitespace-pre-wrap break-words'}, text)\n : h('div', {className:'text-[12px] text-text-muted italic mt-0.5'}, '(no text content)'),\n (m.attachments||[]).length > 0 && h('div', {className:'text-[10px] text-text-muted mt-1 flex items-center gap-1'},\n h(Icon,{name:'attach_file',size:11}), m.attachments.length + ' attachment' + (m.attachments.length===1?'':'s'))));\n })),\n composer);\n}\n`;\n",[],[359,584],"2566a400-5b54-11f1-a375-ee650bb23af1","sheets.ts","d72cca9e-5b4e-11f1-8321-ee650bb23af1",550,["D",1780055238787],"// SheetsRenderer — per-integration renderer for the 'sheets' integration.\nexport const RENDERER = `\nexport function SheetsRenderer({data, endpoint}) {\n if (endpoint === 'values' && Array.isArray(data?.values)) {\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[12px] font-bold'}, data.range),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, data.values.length + ' rows - ' + (data.values[0]?.length || 0) + ' cols')\n ),\n h('div', {className:'overflow-x-auto max-h-[480px]'},\n h('table', {className:'w-full text-[12px]'},\n h('tbody', null,\n data.values.slice(0, 100).map((row, i) => h('tr', {key:i, className:'border-b border-border-subtle/50 hover:bg-white/[0.02]'},\n h('td', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-r border-border-subtle/50', style:{width:40}}, i+1),\n row.map((cell, j) => h('td', {key:j, className:'px-3 py-1.5 ' + (i===0?'font-bold bg-surface-container-lowest':'')}, String(cell)))\n ))\n )\n )\n )\n );\n }\n const sheets = data?.files || [];\n if (!sheets.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No spreadsheets.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(260px, 1fr))'}},\n sheets.map(s => h('a', {\n key:s.id, href:s.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#0f9d5818',color:'#0f9d58'}},\n h(Icon,{name:'table_chart',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, s.name),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted'},\n h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), (s.id||'').slice(0,8) + '-'\n ),\n h('span', null, '-'),\n h('span', null, fmtRelative(s.modifiedTime))\n ),\n s.owners?.[0] && h('div', {className:'text-[10px] text-on-surface-variant truncate mt-1 flex items-center gap-1'},\n h(Icon,{name:'person',size:11}), s.owners[0].displayName\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,592],"d6a99aac-5b4e-11f1-a72a-ee650bb23af1","producthunt.ts","d6a99b74-5b4e-11f1-a72a-ee650bb23af1",532,["D",1780052959207],"// ProductHuntRenderer — per-integration renderer for the 'producthunt' integration.\nexport const RENDERER = `\nexport function ProductHuntRenderer({data, endpoint}) {\n const PH = '#da552f';\n const errs = data?.errors;\n if (Array.isArray(errs) && errs.length) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[13px] mb-2'},\n h(Icon,{name:'error',size:16}), 'Product Hunt API error'),\n h('div', {className:'space-y-1'},\n errs.slice(0,5).map((e,i) => h('div', {key:i, className:'text-[12px] text-on-surface-variant'}, e.message || String(e)))\n )\n );\n }\n\n const topicEdges = data?.data?.topics?.edges;\n if (endpoint === 'topics' || (Array.isArray(topicEdges) && topicEdges.length)) {\n const topics = (topicEdges || []).map(e => e.node || e);\n if (!topics.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No topics.');\n const sorted = [...topics].sort((a,b) => (b.followersCount||0) - (a.followersCount||0));\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-2 mb-3'},\n h(Icon,{name:'tag',size:18,style:{color:PH}}),\n h('h3', {className:'font-bold text-[15px]'}, topics.length + ' Topics')\n ),\n h('div', {className:'flex flex-wrap gap-2'},\n sorted.map(t => h('span', {\n key:t.id || t.slug || t.name,\n className:'inline-flex items-center gap-1.5 px-3 py-1.5 rounded-full text-[12px] font-medium bg-surface-container-high text-on-surface border border-border-subtle'\n },\n t.name,\n t.followersCount != null && h('span', {className:'inline-flex items-center gap-0.5 text-[10px] font-mono-data text-text-muted'},\n h(Icon,{name:'group',size:11}), Number(t.followersCount).toLocaleString()\n )\n ))\n )\n );\n }\n\n const postEdges = data?.data?.posts?.edges || [];\n const posts = postEdges.map(e => e.node || e);\n if (!posts.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No products.');\n\n const chip = (icon, value, color) => h('span', {\n className:'inline-flex items-center gap-1 px-2 py-1 rounded-lg text-[11px] font-bold font-mono-data',\n style:{background:(color||PH)+'18', color:color||PH}\n }, h(Icon,{name:icon,size:12}), Number(value||0).toLocaleString());\n\n return h('div', {className:'mt-3 space-y-2'},\n posts.map((p, idx) => {\n const c = colorFromName(p.name);\n const postTopics = (p.topics?.edges || []).map(e => (e.node||e).name).filter(Boolean).slice(0,3);\n const inner = [\n h('div', {key:'rank', className:'flex flex-col items-center justify-center w-7 flex-shrink-0 text-text-muted font-mono-data text-[13px] font-bold'}, idx + 1),\n p.thumbnail?.url\n ? h('img', {key:'thumb', src:p.thumbnail.url, alt:p.name, className:'w-12 h-12 rounded-lg object-cover flex-shrink-0 border border-border-subtle'})\n : h('div', {key:'thumb', className:'w-12 h-12 rounded-lg flex items-center justify-center font-bold text-[16px] flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(p.name)),\n h('div', {key:'body', className:'flex-1 min-w-0'},\n h('div', {className:'flex items-center gap-2'},\n h('span', {className:'text-[14px] font-bold text-on-surface truncate group-hover:text-secondary transition-colors'}, p.name),\n p.url && h(Icon,{name:'open_in_new',size:13,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n p.tagline && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, p.tagline),\n h('div', {className:'flex items-center gap-1.5 mt-2 flex-wrap'},\n chip('arrow_upward', p.votesCount, PH),\n chip('forum', p.commentsCount, '#94a3b8'),\n postTopics.map(t => h('span', {key:t, className:'px-2 py-0.5 rounded-full text-[10px] text-secondary bg-secondary/10 border border-border-subtle'}, t)),\n p.createdAt && h('span', {className:'text-[10px] text-text-muted font-mono-data ml-auto'}, fmtRelative(p.createdAt))\n )\n )\n ];\n const cls = 'glass-panel rounded-xl p-md flex items-center gap-3 transition-all group ' + (p.url ? 'hover:border-secondary/40 cursor-pointer' : 'border border-border-subtle');\n return p.url\n ? h('a', {key:p.id || idx, href:p.url, target:'_blank', rel:'noreferrer', className:cls}, inner)\n : h('div', {key:p.id || idx, className:cls}, inner);\n })\n );\n}\n`;\n",[],[359,600],"d63bc72a-5b4e-11f1-b9cd-ee650bb23af1","outlook.ts","d63bc806-5b4e-11f1-b9cd-ee650bb23af1",531,["D",1780052958495],"// OutlookRenderer — per-integration renderer for the 'outlook' integration.\nexport const RENDERER = `\nexport function OutlookRenderer({data, endpoint}) {\n const OUTLOOK_BLUE = '#0078d4';\n if (data && data.error) {\n const e = data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40'},\n h('div', {className:'flex items-center gap-2 text-error font-bold text-[14px]'},\n h(Icon,{name:'error',size:18}), 'Outlook error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1.5 font-mono-data break-words'},\n (e.code ? e.code + ': ' : '') + (e.message || String(e)))\n );\n }\n const stripHtml = (s) => (s||'').replace(/<[^>]+>/g,'').replace(/ /g,' ').replace(/&/g,'&').trim();\n\n // --- CONTACTS --------------------------------------------------------------\n if (endpoint === 'contacts') {\n const items = Array.isArray(data?.value) ? data.value : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No contacts.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n items.map((c,i) => {\n const name = c.displayName || [c.givenName,c.surname].filter(Boolean).join(' ') || '(no name)';\n const email = c.emailAddresses?.[0]?.address || '';\n const phone = c.mobilePhone || (c.businessPhones||[])[0] || (c.homePhones||[])[0] || '';\n const col = colorFromName(name);\n return h('div', {key:c.id||i, className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/30 transition-all'},\n h('div', {className:'w-11 h-11 rounded-full flex items-center justify-center font-bold text-[13px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, name),\n (c.jobTitle || c.companyName) && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, [c.jobTitle,c.companyName].filter(Boolean).join(' - ')),\n email && h('a', {href:'mailto:'+email, className:'text-[11px] text-secondary truncate mt-1 flex items-center gap-1 hover:underline'},\n h(Icon,{name:'mail',size:11}), email),\n phone && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1 font-mono-data'},\n h(Icon,{name:'call',size:11}), phone)\n )\n );\n })\n )\n );\n }\n\n // --- CALENDAR --------------------------------------------------------------\n if (endpoint === 'calendar') {\n const events = Array.isArray(data?.value) ? data.value : [];\n if (!events.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No events scheduled.');\n const parse = (dt) => dt?.dateTime ? new Date(dt.dateTime + (dt.timeZone === 'UTC' ? 'Z' : '')) : null;\n const groups = {};\n events.forEach(e => {\n const s = parse(e.start);\n const k = s ? s.toDateString() : 'undated';\n (groups[k] = groups[k] || []).push(e);\n });\n const todayStr = new Date().toDateString();\n const tomStr = new Date(Date.now()+86400000).toDateString();\n const dateLabel = (d) => d === todayStr ? 'TODAY' : d === tomStr ? 'TOMORROW' : new Date(d).toLocaleDateString('en-US',{weekday:'long',month:'short',day:'numeric'}).toUpperCase();\n const fmtT = (dt) => { const x = parse(dt); return x ? x.toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : ''; };\n return h('div', {className:'mt-3 space-y-4'},\n Object.entries(groups).map(([k, evs]) => h('div', {key:k},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-secondary mb-2 sticky top-0 bg-background/80 backdrop-blur-sm py-1'},\n dateLabel(k), h('span', {className:'ml-2 text-text-muted'}, '- ' + evs.length + ' event' + (evs.length===1?'':'s'))),\n h('div', {className:'space-y-2'},\n evs.map((e,i) => {\n const att = (e.attendees||[]).filter(a => a.type !== 'resource');\n const join = e.onlineMeeting?.joinUrl;\n return h('a', {key:e.id||i, href:e.webLink || '#', target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-lg p-3 flex items-start gap-3 hover:border-secondary/30 transition-all'},\n h('div', {className:'flex flex-col items-center justify-center w-16 flex-shrink-0 pr-2 border-r border-border-subtle'},\n h('div', {className:'text-[12px] font-bold text-on-surface font-mono-data'}, e.isAllDay ? 'All-day' : fmtT(e.start)),\n !e.isAllDay && fmtT(e.end) && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, '- ' + fmtT(e.end))),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, e.subject || '(no title)'),\n e.location?.displayName && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1'},\n h(Icon,{name:'place',size:11}), e.location.displayName),\n e.organizer?.emailAddress?.name && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, 'Organizer: ' + e.organizer.emailAddress.name),\n h('div', {className:'flex items-center gap-2 mt-1.5 text-[10px] text-text-muted flex-wrap'},\n att.length > 0 && h('span', {className:'flex items-center gap-1'}, h(Icon,{name:'group',size:11}), att.length + ' attendee' + (att.length===1?'':'s')),\n (e.isOnlineMeeting || join) && h('span', {className:'inline-flex items-center gap-0.5 text-secondary'}, h(Icon,{name:'videocam',size:11}), 'Online'))),\n e.showAs && h('div', {className:'w-1.5 h-1.5 rounded-full mt-2 flex-shrink-0', style:{background: e.showAs==='busy' ? '#ffb4ab' : '#4edea3'}}));\n })))\n )\n );\n }\n\n // --- INBOX / MESSAGES (default) -------------------------------------------\n const messages = Array.isArray(data?.value) ? data.value : [];\n if (!messages.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No messages.');\n const unreadCount = messages.filter(m => m.isRead === false).length;\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex justify-between items-center bg-surface-container-lowest/50'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'},\n h(Icon,{name:'inbox',size:18, style:{color:OUTLOOK_BLUE}}),\n messages.length.toLocaleString() + ' messages'),\n unreadCount > 0 && h('span', {className:'text-[10px] font-mono-data px-2 py-0.5 rounded-full', style:{background:OUTLOOK_BLUE+'22',color:OUTLOOK_BLUE}}, unreadCount + ' unread')),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n messages.map((m,i) => {\n const ea = m.from?.emailAddress || m.sender?.emailAddress || {};\n const fromName = ea.name || ea.address || '(unknown)';\n const unread = m.isRead === false;\n const col = colorFromName(fromName);\n const preview = stripHtml(m.bodyPreview);\n return h('a', {key:m.id||i, href:m.webLink || '#', target:'_blank', rel:'noreferrer',\n className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors ' + (unread ? 'border-l-2' : ''),\n style: unread ? {borderLeftColor:OUTLOOK_BLUE} : null},\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:col+'33',color:col}}, initialsOf(fromName)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className: 'text-[13px] truncate ' + (unread?'font-bold text-on-surface':'text-on-surface-variant')}, fromName),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.receivedDateTime))),\n h('div', {className:'text-[13px] truncate ' + (unread?'font-semibold text-on-surface':'text-on-surface-variant')}, m.subject || '(no subject)'),\n preview && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, preview),\n (m.hasAttachments || m.importance === 'high' || m.flag?.flagStatus === 'flagged') && h('div', {className:'flex flex-wrap gap-1 mt-1.5'},\n m.importance === 'high' && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-error bg-error/10 border border-error/30'}, h(Icon,{name:'priority_high',size:10}), 'HIGH'),\n m.hasAttachments && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-on-surface-variant bg-surface-container-high'}, h(Icon,{name:'attach_file',size:10}), 'ATTACHMENT'),\n m.flag?.flagStatus === 'flagged' && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-tertiary bg-tertiary/10 border border-tertiary/30'}, h(Icon,{name:'flag',size:10}), 'FLAGGED'))\n )\n );\n })\n )\n );\n}\n`;\n",[],[359,608],"d5d5ef7c-5b4e-11f1-95e0-ee650bb23af1","onedrive.ts","d5d5f026-5b4e-11f1-95e0-ee650bb23af1",530,["D",1780052957825],"// OneDriveRenderer — per-integration renderer for the 'onedrive' integration.\nexport const RENDERER = `\nexport function OneDriveRenderer({data, endpoint}) {\n if (data && data.error) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 text-[13px] text-error'}, 'OneDrive error: ' + (data.error.message || data.error.code || String(data.error)));\n // createFolder returns the new driveItem (single object) → confirmation card.\n if (endpoint === 'createFolder' && data && data.id && data.name) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-4'},\n h('div', {className:'w-12 h-12 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#0078d418',color:'#0078d4'}}, h(Icon,{name:'create_new_folder',size:24})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, 'Folder created: ' + data.name),\n data.webUrl && h('a', {href:data.webUrl, target:'_blank', rel:'noopener', className:'text-[11px] text-primary truncate inline-flex items-center gap-1 mt-0.5'}, h(Icon,{name:'open_in_new',size:11}), 'Open in OneDrive')\n )\n );\n }\n const items = (data && data.value) || [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No files.');\n const mimeIcon = (mime) => {\n if (!mime) return {icon:'insert_drive_file', color:'#94a3b8'};\n if (mime.indexOf('pdf') !== -1) return {icon:'picture_as_pdf', color:'#ea4335'};\n if (mime.indexOf('word') !== -1 || mime.indexOf('document') !== -1) return {icon:'description', color:'#2b579a'};\n if (mime.indexOf('sheet') !== -1 || mime.indexOf('excel') !== -1 || mime.indexOf('csv') !== -1) return {icon:'table_chart', color:'#217346'};\n if (mime.indexOf('presentation') !== -1 || mime.indexOf('powerpoint') !== -1) return {icon:'co_present', color:'#d24726'};\n if (mime.indexOf('zip') !== -1 || mime.indexOf('compressed') !== -1) return {icon:'folder_zip', color:'#5f6368'};\n if (mime.indexOf('image/') === 0) return {icon:'image', color:'#34a853'};\n if (mime.indexOf('video/') === 0) return {icon:'movie', color:'#ea4335'};\n if (mime.indexOf('audio/') === 0) return {icon:'audiotrack', color:'#a078ff'};\n return {icon:'insert_drive_file', color:'#94a3b8'};\n };\n return h('div', {className:'mt-3'},\n h('div', {className:'flex items-center justify-between mb-3'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'}, h(Icon,{name:'cloud',size:18,className:'text-[#0078d4]'}), items.length + (items.length === 1 ? ' item' : ' items')),\n h('span', {className:'text-[10px] font-mono-data text-text-muted uppercase tracking-widest'}, endpoint || 'recent')\n ),\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(220px, 1fr))'}},\n items.map(f => {\n const isFolder = !!f.folder;\n const ico = isFolder ? {icon:'folder', color:'#ffc83d'} : mimeIcon(f.file && f.file.mimeType);\n const ownerName = (f.createdBy && f.createdBy.user && f.createdBy.user.displayName) || (f.lastModifiedBy && f.lastModifiedBy.user && f.lastModifiedBy.user.displayName) || '-';\n const c = colorFromName(ownerName);\n const subtitle = isFolder ? ((f.folder.childCount != null ? f.folder.childCount + ' items' : 'Folder')) : ((f.file && f.file.mimeType) ? f.file.mimeType.replace(/^application\\\\//,'').replace(/^vnd\\\\.openxmlformats-officedocument\\\\./,'') : 'File');\n return h('a', {\n key: f.id, href: f.webUrl || '#', target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex flex-col gap-2 hover:border-primary/30 transition-all group'\n },\n h('div', {className:'flex items-start justify-between gap-2'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background: ico.color+'18', color: ico.color}}, h(Icon,{name: ico.icon, size:22})),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, f.name),\n h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, subtitle)\n ),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted pt-2 border-t border-border-subtle'},\n h('span', {className:'truncate flex items-center gap-1'},\n h('span', {className:'w-4 h-4 rounded-full flex items-center justify-center text-[7px] font-bold flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(ownerName)),\n ownerName\n ),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(f.lastModifiedDateTime))\n ),\n (!isFolder && f.size) ? h('div', {className:'flex justify-end'}, h('span', {className:'font-mono-data text-[9px] text-on-surface-variant'}, fmtFileSize(f.size))) : (isFolder ? h('div', {className:'flex justify-end'}, h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-[#0078d4]/10 text-[#0078d4]'}, h(Icon,{name:'folder',size:9}), 'FOLDER')) : null)\n );\n })\n )\n );\n}\n`;\n",[],[359,616],"d57a019e-5b4e-11f1-89bf-ee650bb23af1","linkedin.ts","d57a0248-5b4e-11f1-89bf-ee650bb23af1",529,["D",1780052957225],"// LinkedInRenderer — per-integration renderer for the 'linkedin' integration.\nexport const RENDERER = `\nexport function LinkedInRenderer({data, endpoint}) {\n const {useState} = React;\n const LI_BLUE = '#0077b5';\n const LI_LIGHT = '#00a0dc';\n\n // Local tab: profile | posts | analytics | compose\n const [tab, setTab] = useState(endpoint === 'analytics' ? 'analytics' : endpoint === 'posts' ? 'posts' : 'profile');\n const [draft, setDraft] = useState('');\n const [posting, setPosting] = useState(false);\n const [posted, setPosted] = useState('');\n\n // -- helpers --------------------------------------------------------------\n const fmtDate = (ms) => {\n if (!ms) return '-';\n try { return new Date(Number(ms)).toLocaleDateString(undefined, {month:'short', day:'numeric', year:'numeric'}); }\n catch { return '-'; }\n };\n\n const lifecycleBadge = (state) => {\n const color = state === 'PUBLISHED' ? '#4edea3' : state === 'DRAFT' ? '#fbbc04' : '#94a3b8';\n return h('span', {style:{background:color+'22',color,border:'1px solid '+color+'66',borderRadius:4,padding:'1px 7px',fontSize:10,fontWeight:700,letterSpacing:.5}}, state || '-');\n };\n\n // -- tab bar --------------------------------------------------------------\n const tabBtn = (id, label) => h('button', {\n key:id,\n onClick:() => setTab(id),\n style:{\n padding:'5px 16px', borderRadius:20, fontSize:12, fontWeight:700, cursor:'pointer',\n background: tab===id ? LI_BLUE : 'transparent',\n color: tab===id ? '#fff' : LI_BLUE,\n border: '1px solid '+LI_BLUE,\n transition:'all .15s'\n }\n }, label);\n\n const tabRow = h('div', {style:{display:'flex',gap:8,marginBottom:16,flexWrap:'wrap'}},\n tabBtn('profile','Profile'),\n tabBtn('posts','Posts'),\n tabBtn('analytics','Analytics'),\n tabBtn('compose','Compose')\n );\n\n // -- compose view: draft a post and queue it for approval -----------------\n const composeView = () => {\n const submit = async () => {\n const text = draft.trim();\n if (!text) return;\n setPosting(true); setPosted('');\n try {\n const r = await fetch('/api/actions', {\n method:'POST', headers:{'Content-Type':'application/json'},\n body: JSON.stringify({ integration:'linkedin', endpoint:'create-post', source_type:'linkedin', workspace:'sales', title:'LinkedIn post', preview: text.slice(0,140), payload:{ text } })\n });\n if (r.ok) { setPosted('Queued in Approvals for one-click publish.'); setDraft(''); }\n else { const j = await r.json().catch(() => ({})); setPosted('Error: ' + (j.error || ('HTTP ' + r.status))); }\n } catch (e) { setPosted('Connection error'); }\n setPosting(false);\n };\n return h('div', {className:'mt-3'},\n h('div', {className:'glass-panel rounded-xl p-lg space-y-3'},\n h('div', {className:'flex items-center gap-2'},\n h('div', {className:'w-8 h-8 rounded-full flex items-center justify-center text-[12px] font-bold', style:{background:'linear-gradient(135deg,'+LI_BLUE+','+LI_LIGHT+')', color:'#fff'}}, 'in'),\n h('span', {className:'text-[13px] font-bold text-on-surface'}, 'Compose LinkedIn post')\n ),\n h('textarea', {\n value: draft, onChange: e => setDraft(e.target.value), rows:6, maxLength:3000,\n placeholder:'Share an update on LinkedIn...',\n className:'w-full bg-surface-container-lowest border border-border-subtle rounded-lg px-3 py-2.5 text-[13px] text-on-surface outline-none focus:border-primary resize-none'\n }),\n h('div', {className:'flex items-center justify-between gap-3'},\n h('span', {className:'text-[11px] text-text-muted font-mono-data'}, draft.length + ' / 3000'),\n h('button', {\n onClick: submit, disabled: posting || !draft.trim(),\n className:'px-4 py-2 rounded-lg text-white text-[13px] font-semibold disabled:opacity-50',\n style:{background:LI_BLUE}\n }, posting ? 'Queuing...' : 'Queue post')\n ),\n posted && h('div', {className:'text-[12px] ' + (posted.startsWith('Queued') ? 'text-secondary' : 'text-error')}, posted),\n h('div', {className:'text-[11px] text-text-muted'}, 'Nothing is posted to LinkedIn until you approve it in Approvals.')\n )\n );\n };\n\n // -- profile view ---------------------------------------------------------\n const profileView = () => {\n const name = data?.name || [data?.localizedFirstName, data?.localizedLastName].filter(Boolean).join(' ') || '-';\n const head = data?.headline || data?.localizedHeadline || '';\n const lid = data?.id || '';\n const c = colorFromName(name);\n return h('div', {className:'glass-panel rounded-xl p-lg flex items-center gap-4 mt-3'},\n h('div', {\n className:'w-14 h-14 rounded-full flex items-center justify-center font-bold text-[18px] flex-shrink-0',\n style:{background:'linear-gradient(135deg,'+LI_BLUE+','+LI_LIGHT+')', color:'#fff'}\n }, initialsOf(name)),\n h('div', {className:'flex-1 min-w-0'},\n h('h3', {className:'font-bold text-[18px] text-on-surface'}, name),\n head && h('div', {className:'text-[13px] text-on-surface-variant mt-0.5'}, head),\n lid && h('div', {className:'text-[11px] text-text-muted font-mono-data mt-1'}, 'ID: ' + lid)\n )\n );\n };\n\n // -- posts view -----------------------------------------------------------\n const postsView = () => {\n const items = Array.isArray(data?.elements) ? data.elements : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No posts found.');\n return h('div', {className:'mt-3 space-y-2 max-h-[520px] overflow-y-auto'},\n items.map((p,i) => {\n const urn = (p.id || p.urn || '').toString();\n const ts = p.created?.time;\n const state = p.lifecycleState || p.visibility || '';\n return h('div', {key:i, className:'glass-panel rounded-xl p-md border border-border-subtle'},\n h('div', {className:'flex items-center justify-between gap-3 flex-wrap'},\n h('span', {className:'text-[12px] text-text-muted font-mono-data'}, fmtDate(ts)),\n lifecycleBadge(state)\n ),\n urn && h('div', {className:'flex items-center justify-between gap-2 mt-1.5'},\n h('span', {className:'text-[11px] text-text-muted font-mono-data truncate'}, urn.length > 50 ? urn.slice(0,50)+'...' : urn),\n h('a', {href:'https://www.linkedin.com/feed/update/'+encodeURIComponent(urn), target:'_blank', rel:'noreferrer', className:'text-[11px] font-semibold flex items-center gap-1 flex-shrink-0', style:{color:LI_BLUE}}, h(Icon,{name:'open_in_new',size:12}), 'Open')\n )\n );\n })\n );\n };\n\n // -- analytics view -------------------------------------------------------\n const analyticsView = () => {\n const items = Array.isArray(data?.elements) ? data.elements : [];\n if (!items.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No analytics data.');\n const totals = items.reduce((acc, el) => {\n const s = el.shareStatistics || {};\n acc.impressions += (s.impressionCount || 0);\n acc.clicks += (s.clickCount || 0);\n acc.likes += (s.likeCount || 0);\n acc.comments += (s.commentCount || 0);\n return acc;\n }, {impressions:0,clicks:0,likes:0,comments:0});\n const chip = (label, value, icon) => h('div', {className:'glass-panel rounded-xl p-md flex flex-col items-center gap-1 border border-border-subtle min-w-[110px]'},\n h('span', {className:'text-[24px] font-bold text-on-surface'}, value.toLocaleString()),\n h('span', {className:'text-[11px] text-text-muted uppercase tracking-widest'}, label)\n );\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(110px, 1fr))'}},\n chip('Impressions', totals.impressions),\n chip('Clicks', totals.clicks),\n chip('Likes', totals.likes),\n chip('Comments', totals.comments)\n ),\n h('div', {className:'text-[11px] text-text-muted mt-3 text-right'}, items.length + ' post(s) aggregated')\n );\n };\n\n return h('div', {className:'text-on-surface'},\n tabRow,\n tab === 'profile' ? profileView() :\n tab === 'posts' ? postsView() :\n tab === 'compose' ? composeView() :\n analyticsView()\n );\n}\n`;\n",[],[359,624],"d50ecbe0-5b4e-11f1-89bf-ee650bb23af1","gsc.ts","d50eccb2-5b4e-11f1-89bf-ee650bb23af1",528,["D",1780052956519],"// GSCRenderer — per-integration renderer for the 'gsc' integration.\nexport const RENDERER = `\nexport function GSCRenderer({data, endpoint}) {\n if (data && data.error) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error mb-1'}, 'Search Console error'),\n h('div', {className:'text-[12px] text-on-surface-variant leading-relaxed'}, String((data.error && data.error.message) || data.error))\n )\n );\n }\n\n const fmtNum = (n) => (n == null ? '-' : Math.round(n).toLocaleString());\n const fmtPct = (r) => (r == null ? '-' : (r * 100).toFixed(1) + '%');\n const fmtPos = (p) => (p == null ? '-' : p.toFixed(1));\n\n // --- SITEMAPS ---------------------------------------------------------------\n if (endpoint === 'sitemaps') {\n const maps = Array.isArray(data && data.sitemap) ? data.sitemap : [];\n if (!maps.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No sitemaps submitted.');\n return h('div', {className:'mt-3 space-y-2'},\n maps.map((s, i) => {\n const warn = parseInt(s.warnings || 0, 10);\n const err = parseInt(s.errors || 0, 10);\n const ok = !err && !warn && !s.isPending;\n const accent = err ? '#ffb4ab' : warn ? '#fbbc04' : s.isPending ? '#94a3b8' : '#4edea3';\n return h('a', {\n key:i, href:s.path, target:'_blank', rel:'noreferrer',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:accent+'18', color:accent}},\n h(Icon, {name:'account_tree', size:22})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, (s.path || '').replace(/^https?:\\\\/\\\\//,'')),\n h('div', {className:'flex flex-wrap items-center gap-2 mt-1.5 text-[10px]'},\n s.type && h('span', {className:'px-1.5 py-0.5 rounded bg-surface-container-high text-on-surface-variant uppercase font-semibold'}, s.type),\n s.lastSubmitted && h('span', {className:'text-text-muted font-mono-data'}, 'submitted ' + fmtRelative(s.lastSubmitted)),\n s.isPending && h('span', {className:'px-1.5 py-0.5 rounded bg-surface-container-high text-text-muted font-semibold'}, 'PENDING'),\n err > 0 && h('span', {className:'px-1.5 py-0.5 rounded text-error font-semibold', style:{background:'#ffb4ab18'}}, err + ' error' + (err===1?'':'s')),\n warn > 0 && h('span', {className:'px-1.5 py-0.5 rounded text-tertiary font-semibold', style:{background:'#fbbc0418'}}, warn + ' warning' + (warn===1?'':'s')),\n ok && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-secondary font-semibold', style:{background:'#4edea318'}}, h(Icon,{name:'check_circle',size:10,filled:true}), 'HEALTHY'),\n Array.isArray(s.contents) && s.contents.length > 0 && h('span', {className:'text-text-muted font-mono-data'}, s.contents.length + ' content type' + (s.contents.length===1?'':'s'))\n )\n ),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n );\n })\n );\n }\n\n // --- SEARCH ANALYTICS / PAGES -----------------------------------------------\n const isPages = endpoint === 'pages';\n const rows = Array.isArray(data && data.rows) ? data.rows : [];\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No search data for this range.');\n\n const sorted = rows.slice().sort((a, b) => (b.clicks || 0) - (a.clicks || 0));\n const maxClicks = Math.max(1, ...sorted.map(r => r.clicks || 0));\n const totals = sorted.reduce((acc, r) => {\n acc.clicks += (r.clicks || 0);\n acc.impressions += (r.impressions || 0);\n return acc;\n }, {clicks:0, impressions:0});\n const avgCtr = totals.impressions ? totals.clicks / totals.impressions : 0;\n const avgPos = sorted.reduce((s, r) => s + (r.position || 0), 0) / sorted.length;\n\n const chip = (label, value, icon, color) => h('div', {className:'glass-panel rounded-xl p-md flex flex-col gap-1'},\n h('div', {className:'flex items-center gap-1.5 text-[10px] uppercase tracking-widest text-text-muted'}, h(Icon,{name:icon,size:12,className:'text-secondary'}), label),\n h('div', {className:'text-[22px] font-bold font-mono-data', style:color?{color}:null}, value)\n );\n\n return h('div', {className:'mt-3 space-y-4'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(140px, 1fr))'}},\n chip('Total Clicks', fmtNum(totals.clicks), 'ads_click'),\n chip('Impressions', fmtNum(totals.impressions), 'visibility'),\n chip('Avg CTR', fmtPct(avgCtr), 'percent'),\n chip('Avg Position', fmtPos(avgPos), 'leaderboard')\n ),\n h('div', {className:'glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'},\n h(Icon,{name: isPages ? 'description' : 'search', size:16, className:'text-secondary'}),\n isPages ? 'Top Pages' : 'Top Queries'\n ),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, sorted.length + (isPages ? ' pages' : ' queries'))\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[520px] overflow-y-auto'},\n sorted.map((r, i) => {\n const key = (r.keys && r.keys[0]) || '';\n const pct = ((r.clicks || 0) / maxClicks) * 100;\n const label = isPages ? key.replace(/^https?:\\\\/\\\\//,'') : key;\n const inner = h('div', {className:'flex items-center gap-3 p-3 hover:bg-white/[0.02] transition-colors' + (isPages ? ' cursor-pointer group' : '')},\n h('div', {className:'w-6 text-center text-[11px] font-mono-data text-text-muted flex-shrink-0'}, i + 1),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-semibold text-on-surface truncate flex items-center gap-1.5'},\n label || '(empty)',\n isPages && h(Icon, {name:'open_in_new', size:12, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n h('div', {className:'h-1.5 mt-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full transition-all', style:{width: Math.max(2, pct) + '%', background:'linear-gradient(90deg,#4edea3,#6ffbbe)'}})\n )\n ),\n h('div', {className:'flex items-center gap-4 flex-shrink-0 text-right'},\n h('div', {className:'w-14'},\n h('div', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtNum(r.clicks)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'clicks')\n ),\n h('div', {className:'w-16 hidden sm:block'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface-variant'}, fmtNum(r.impressions)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'impr')\n ),\n h('div', {className:'w-12'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface'}, fmtPct(r.ctr)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'ctr')\n ),\n h('div', {className:'w-12'},\n h('div', {className:'text-[12px] font-mono-data text-on-surface'}, fmtPos(r.position)),\n h('div', {className:'text-[9px] text-text-muted uppercase'}, 'pos')\n )\n )\n );\n return isPages && key\n ? h('a', {key:i, href:key, target:'_blank', rel:'noreferrer', className:'block'}, inner)\n : h('div', {key:i}, inner);\n })\n )\n )\n );\n}\n`;\n",[],[359,632],"d4ae5f44-5b4e-11f1-a491-ee650bb23af1","gmail.ts","d4ae603e-5b4e-11f1-a491-ee650bb23af1",527,["D",1780052955889],"// GmailRenderer — per-integration renderer for the 'gmail' integration.\nexport const RENDERER = `\nexport function GmailRenderer({data, endpoint}) {\n if (endpoint === 'profile' && data?.emailAddress) {\n return h('div', {className:'glass-panel rounded-xl p-lg flex items-center gap-4 mt-3'},\n h('div', {\n className:'w-14 h-14 rounded-full flex items-center justify-center text-on-primary font-bold text-[18px]',\n style:{background:'linear-gradient(135deg,#ea4335,#c0c1ff)'}\n }, initialsOf(data.emailAddress)),\n h('div', {className:'flex-1'},\n h('h3', {className:'font-bold text-[18px]'}, data.emailAddress),\n h('div', {className:'flex gap-4 mt-1 text-[12px] text-text-muted font-mono-data'},\n h('span', null, h(Icon,{name:'mail',size:12,className:'inline mr-1'}), data.messagesTotal?.toLocaleString() + ' messages'),\n h('span', null, h(Icon,{name:'forum',size:12,className:'inline mr-1'}), data.threadsTotal?.toLocaleString() + ' threads'),\n h('span', null, 'History ID: ' + data.historyId)\n )\n )\n );\n }\n if (endpoint === 'labels' && Array.isArray(data?.labels)) {\n const SYSTEM = data.labels.filter(l => l.type === 'system');\n const USER = data.labels.filter(l => l.type !== 'system');\n const chip = (l) => h('span', {key:l.id, className:'px-2.5 py-1 rounded-full text-[11px] font-medium bg-surface-container-high text-on-surface-variant border border-border-subtle'},\n l.name, ' ', l.messagesUnread > 0 && h('span', {className:'ml-1 px-1.5 rounded bg-primary/20 text-primary text-[10px]'}, l.messagesUnread)\n );\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 space-y-md'},\n h('div', null,\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-2'}, 'System Labels'),\n h('div', {className:'flex flex-wrap gap-1.5'}, SYSTEM.map(chip))\n ),\n USER.length > 0 && h('div', null,\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-2'}, 'Your Labels'),\n h('div', {className:'flex flex-wrap gap-1.5'}, USER.map(chip))\n )\n );\n }\n const messages = Array.isArray(data?._hydrated) ? data._hydrated : [];\n if (!messages.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No messages.');\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-4 border-b border-border-subtle flex justify-between items-center bg-surface-container-lowest/50'},\n h('h3', {className:'text-[14px] font-bold flex items-center gap-2'},\n h(Icon,{name:'inbox',size:18,className:'text-error'}),\n (data.resultSizeEstimate || messages.length).toLocaleString() + ' messages'\n ),\n h('span', {className:'text-[10px] font-mono-data text-text-muted'}, 'showing ' + messages.length)\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n messages.map(m => {\n const fromName = (m.from || '').replace(/<.*>/,'').trim() || m.from;\n const c = colorFromName(fromName);\n return h('div', {\n key: m.id,\n className:'flex items-start gap-3 p-3.5 hover:bg-white/[0.02] transition-colors cursor-pointer ' + (m.unread ? 'border-l-2 border-l-primary' : '')\n },\n h('div', {className:'w-9 h-9 rounded-full flex items-center justify-center font-bold text-[11px] flex-shrink-0', style:{background:c+'33',color:c}}, initialsOf(fromName)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'flex justify-between items-baseline gap-3'},\n h('span', {className: 'text-[13px] truncate ' + (m.unread?'font-bold text-on-surface':'text-on-surface-variant')}, fromName),\n h('span', {className:'text-[10px] text-text-muted font-mono-data flex-shrink-0'}, fmtRelative(m.internalDate ? new Date(parseInt(m.internalDate,10)).toISOString() : m.date))\n ),\n h('div', {className:'text-[13px] truncate ' + (m.unread?'font-semibold text-on-surface':'text-on-surface-variant')}, m.subject || '(no subject)'),\n m.snippet && h('div', {className:'text-[12px] text-text-muted truncate mt-0.5'}, m.snippet),\n (m.starred || (m.labelIds||[]).filter(l => !['INBOX','UNREAD','CATEGORY_PERSONAL','SENT','STARRED'].includes(l)).length > 0) && h('div', {className:'flex flex-wrap gap-1 mt-1.5'},\n m.starred && h('span', {className:'inline-flex items-center gap-0.5 px-1.5 py-0.5 rounded text-[9px] font-semibold text-tertiary bg-tertiary/10 border border-tertiary/30'},\n h(Icon,{name:'star',size:10,filled:true}), 'STARRED'\n ),\n (m.labelIds||[]).filter(l => !['INBOX','UNREAD','CATEGORY_PERSONAL','SENT','STARRED'].includes(l)).slice(0,3).map(l =>\n h('span', {key:l, className:'px-1.5 py-0.5 rounded text-[9px] font-semibold text-on-surface-variant bg-surface-container-high'}, l.replace('CATEGORY_','').replace(/_/g,' '))\n )\n )\n )\n );\n })\n )\n );\n}\n\n// --- DRIVE -------------------------------------------------------------------\nconst MIME_ICON = {\n 'application/vnd.google-apps.folder': {icon:'folder', color:'#fbbc04'},\n 'application/vnd.google-apps.document': {icon:'description', color:'#4285f4'},\n 'application/vnd.google-apps.spreadsheet': {icon:'table_chart', color:'#0f9d58'},\n 'application/vnd.google-apps.presentation': {icon:'co_present', color:'#f4b400'},\n 'application/vnd.google-apps.form': {icon:'list_alt', color:'#673ab7'},\n 'application/pdf': {icon:'picture_as_pdf', color:'#ea4335'},\n 'application/zip': {icon:'folder_zip', color:'#5f6368'},\n 'image/png': {icon:'image', color:'#34a853'},\n 'image/jpeg': {icon:'image', color:'#34a853'},\n 'video/mp4': {icon:'movie', color:'#ea4335'},\n};\nconst driveIconFor = (mime) => MIME_ICON[mime] || (mime?.startsWith('image/') ? MIME_ICON['image/png'] : mime?.startsWith('video/') ? MIME_ICON['video/mp4'] : {icon:'insert_drive_file', color:'#94a3b8'});\n`;\n",[],[359,640],"d43df1aa-5b4e-11f1-90b4-ee650bb23af1","ga4.ts","d43df27c-5b4e-11f1-90b4-ee650bb23af1",526,["D",1780052955152],"// GA4Renderer — per-integration renderer for the 'ga4' integration.\nexport const RENDERER = `\nexport function GA4Renderer({data, endpoint}) {\n if (data && data.error) {\n const msg = (data.error && data.error.message) || data.error;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon, {name:'error', size:20, className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error mb-1'}, 'Google Analytics error'),\n h('div', {className:'text-[12px] text-on-surface-variant leading-relaxed'}, String(msg))\n )\n );\n }\n const fmtN = (v) => { const n = parseFloat(v) || 0; return n >= 1e6 ? (n/1e6).toFixed(1)+'M' : n >= 1e3 ? (n/1e3).toFixed(1)+'K' : String(Math.round(n*100)/100); };\n const isRate = (k) => /rate|percent/i.test(k || '');\n const fmtMetric = (k, v) => isRate(k) ? (parseFloat(v||'0')*100).toFixed(1)+'%' : fmtN(v);\n const parseRows = (d) => {\n if (!d || !d.rows || !d.rows.length) return [];\n const dh = (d.dimensionHeaders || []).map(x => x.name);\n const mh = (d.metricHeaders || []).map(x => x.name);\n return d.rows.map(r => {\n const o = {};\n dh.forEach((k,i) => { o[k] = (r.dimensionValues || [])[i] && r.dimensionValues[i].value; });\n mh.forEach((k,i) => { o[k] = (r.metricValues || [])[i] && r.metricValues[i].value; });\n return o;\n });\n };\n const empty = (t) => h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, t);\n const dimNames = (data && data.dimensionHeaders || []).map(x => x.name);\n const metNames = (data && data.metricHeaders || []).map(x => x.name);\n\n // --- KPI (two date ranges; current then previous metrics) -------------------\n if (endpoint === 'kpi') {\n const row = data && data.rows && data.rows[0];\n if (!row || !metNames.length) return empty('No KPI data for this range.');\n const vals = row.metricValues || [];\n const n = metNames.length;\n const META = {sessions:'#c0c1ff', activeUsers:'#4edea3', newUsers:'#d0bcff', screenPageViews:'#8083ff', engagementRate:'#6ffbbe', bounceRate:'#ffb4ab'};\n const labelOf = (k) => k.replace(/([A-Z])/g,' $1').replace(/^./, c => c.toUpperCase());\n return h('div', {className:'grid gap-3 mt-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(155px, 1fr))'}},\n metNames.map((k, idx) => {\n const color = META[k] || '#c0c1ff';\n const cur = parseFloat((vals[idx] && vals[idx].value) || '0');\n const prev = (idx + n) < vals.length ? parseFloat((vals[idx+n] && vals[idx+n].value) || '0') : null;\n const delta = (prev != null && prev !== 0) ? ((cur - prev) / prev) * 100 : null;\n const up = delta != null && delta >= 0;\n const good = isRate(k) ? (/bounce/i.test(k) ? !up : up) : up;\n return h('div', {key:k, className:'glass-panel rounded-xl p-4', style:{borderColor:color+'28'}},\n h('div', {className:'text-[10px] text-text-muted uppercase tracking-wider mb-2'}, labelOf(k)),\n h('div', {className:'flex items-baseline gap-2 flex-wrap'},\n h('div', {className:'text-[26px] font-extrabold text-on-surface leading-none font-mono-data'}, fmtMetric(k, String(cur))),\n delta != null && h('span', {className:'inline-flex items-center gap-0.5 text-[11px] font-bold', style:{color: good ? '#4edea3' : '#ffb4ab'}},\n h(Icon, {name: up ? 'trending_up' : 'trending_down', size:12}), Math.abs(delta).toFixed(0)+'%')\n ),\n h('div', {className:'mt-2 h-0.5 rounded', style:{background:color, opacity:0.35}})\n );\n })\n );\n }\n\n // --- TRAFFIC OVER TIME (date dimension trend table) -------------------------\n if (endpoint === 'traffic-over-time') {\n const rows = parseRows(data);\n if (!rows.length) return empty('No traffic data for this range.');\n const max = Math.max(1, ...rows.map(r => parseFloat(r.sessions || 0)));\n const total = rows.reduce((s,r) => s + (parseFloat(r.sessions)||0), 0);\n return h('div', {className:'mt-3 glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:'show_chart',size:16,className:'text-secondary'}), 'Sessions Over Time'),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, fmtN(total)+' total')\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[440px] overflow-y-auto'},\n rows.map((r,i) => {\n const d = String(r.date || '');\n const lbl = d.length === 8 ? d.slice(4,6)+'/'+d.slice(6,8) : d;\n const pct = (parseFloat(r.sessions||0) / max) * 100;\n return h('div', {key:i, className:'flex items-center gap-3 px-3 py-1.5'},\n h('div', {className:'w-12 text-[11px] font-mono-data text-text-muted flex-shrink-0'}, lbl),\n h('div', {className:'flex-1 h-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#c0c1ff,#8083ff)'}})),\n h('div', {className:'w-14 text-right text-[12px] font-bold font-mono-data text-on-surface'}, fmtN(r.sessions)),\n h('div', {className:'w-12 text-right text-[11px] font-mono-data text-text-muted'}, fmtN(r.activeUsers))\n );\n })\n )\n );\n }\n\n // --- REALTIME (active users by country, last 30 min) ------------------------\n if (endpoint === 'realtime') {\n const rows = parseRows(data);\n const total = rows.reduce((s,r) => s + (parseFloat(r.activeUsers)||0), 0);\n return h('div', {className:'mt-3 space-y-3'},\n h('div', {className:'glass-panel rounded-xl p-4 flex items-center gap-3', style:{borderColor:'#4edea328'}},\n h('span', {className:'w-2.5 h-2.5 rounded-full animate-pulse', style:{background:'#4edea3'}}),\n h('div', {className:'text-[30px] font-extrabold font-mono-data text-on-surface leading-none'}, fmtN(total)),\n h('div', {className:'text-[11px] text-text-muted uppercase tracking-wider'}, 'active users now')\n ),\n rows.length ? h('div', {className:'glass-panel rounded-xl divide-y divide-border-subtle/50'},\n rows.map((r,i) => h('div', {key:i, className:'flex items-center justify-between px-3 py-2'},\n h('span', {className:'text-[13px] text-on-surface'}, r[dimNames[0]] || '(unknown)'),\n h('span', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtN(r.activeUsers))\n ))\n ) : empty('No active users in the last 30 minutes.')\n );\n }\n\n // --- FUNNEL (alpha runFunnelReport) -----------------------------------------\n if (endpoint === 'funnel') {\n const ft = data && (data.funnelTable || data.subFunnelTables && data.subFunnelTables[0]);\n const frows = ft && ft.rows || [];\n if (!frows.length) return empty('No funnel data. Verify event names / property ID.');\n const cnt = (r) => parseFloat((r.metricValues && r.metricValues[0] && r.metricValues[0].value) || 0);\n const dv = (r) => ((r.dimensionValues || []).map(x => x.value).join(' / ')) || 'Step';\n const top = Math.max(1, ...frows.map(cnt));\n return h('div', {className:'mt-3 space-y-2'},\n frows.map((r,i) => {\n const c = cnt(r), pct = (c/top)*100;\n return h('div', {key:i, className:'glass-panel rounded-xl p-3'},\n h('div', {className:'flex justify-between items-center mb-1.5'},\n h('span', {className:'text-[13px] font-semibold text-on-surface'}, (i+1)+'. '+dv(r)),\n h('span', {className:'text-[13px] font-bold font-mono-data text-secondary'}, fmtN(c)+' ('+pct.toFixed(0)+'%)')),\n h('div', {className:'h-2 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#4edea3,#6ffbbe)'}}))\n );\n })\n );\n }\n\n // --- RANKED / TABULAR: user-acquisition, report, page-views, top-events, device-breakdown\n const rows = parseRows(data);\n if (!rows.length) return empty('No data for this range.');\n const isPages = endpoint === 'page-views';\n const sortKey = metNames.find(k => /sessions|eventCount|screenPageViews/i.test(k)) || metNames[0];\n const sorted = rows.slice().sort((a,b) => (parseFloat(b[sortKey])||0) - (parseFloat(a[sortKey])||0));\n const max = Math.max(1, ...sorted.map(r => parseFloat(r[sortKey])||0));\n const titleMap = {'user-acquisition':'Acquisition by Channel','report':'Traffic Report','page-views':'Top Pages','top-events':'Top Events','device-breakdown':'Devices'};\n const iconMap = {'user-acquisition':'campaign','report':'table_chart','page-views':'description','top-events':'bolt','device-breakdown':'devices'};\n const totals = {};\n metNames.forEach(k => { totals[k] = sorted.reduce((s,r) => s + (parseFloat(r[k])||0), 0); });\n return h('div', {className:'mt-3 space-y-4'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(140px, 1fr))'}},\n metNames.slice(0,4).map(k => h('div', {key:k, className:'glass-panel rounded-xl p-md flex flex-col gap-1'},\n h('div', {className:'text-[10px] uppercase tracking-widest text-text-muted'}, k.replace(/([A-Z])/g,' $1')),\n h('div', {className:'text-[22px] font-bold font-mono-data text-on-surface'}, isRate(k) ? (totals[k]/sorted.length*100).toFixed(1)+'%' : fmtN(totals[k]))\n ))\n ),\n h('div', {className:'glass-panel rounded-xl overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle flex justify-between items-center'},\n h('h3', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:iconMap[endpoint]||'analytics',size:16,className:'text-secondary'}), titleMap[endpoint] || 'Report'),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, sorted.length + ' rows')\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[520px] overflow-y-auto'},\n sorted.map((r,i) => {\n const key = dimNames.map(d => r[d]).filter(Boolean).join(' / ') || '(empty)';\n const label = isPages ? key.replace(/^https?:\\\\/\\\\//,'') : key;\n const pct = (parseFloat(r[sortKey])||0) / max * 100;\n const inner = h('div', {className:'flex items-center gap-3 p-3 hover:bg-white/[0.02] transition-colors' + (isPages ? ' group cursor-pointer' : '')},\n h('div', {className:'w-6 text-center text-[11px] font-mono-data text-text-muted flex-shrink-0'}, i+1),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-semibold text-on-surface truncate flex items-center gap-1.5'}, label,\n isPages && h(Icon,{name:'open_in_new',size:12,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})),\n h('div', {className:'h-1.5 mt-1.5 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full rounded-full', style:{width:Math.max(2,pct)+'%', background:'linear-gradient(90deg,#c0c1ff,#8083ff)'}}))\n ),\n h('div', {className:'flex items-center gap-4 flex-shrink-0 text-right'},\n metNames.slice(0,3).map(k => h('div', {key:k, className:'w-14'},\n h('div', {className:'text-[12px] font-bold font-mono-data text-on-surface'}, fmtMetric(k, r[k])),\n h('div', {className:'text-[9px] text-text-muted uppercase truncate'}, k.replace(/([A-Z]).*/,'').slice(0,6) || k.slice(0,6))\n ))\n )\n );\n return isPages && r[dimNames[0]] ? h('a', {key:i, href:r[dimNames[0]], target:'_blank', rel:'noreferrer', className:'block'}, inner) : h('div', {key:i}, inner);\n })\n )\n )\n );\n}\n`;\n",[],[359,648],"d392f818-5b4e-11f1-b8d2-ee650bb23af1","excel.ts","d392f8c2-5b4e-11f1-b8d2-ee650bb23af1",525,["D",1780052954028],"// ExcelRenderer — per-integration renderer for the 'excel' integration.\nexport const RENDERER = `\nexport function ExcelRenderer({data, endpoint}) {\n const GREEN = '#217346';\n if (data && data.error) {\n const e = data.error;\n const msg = (e && typeof e === 'object') ? (e.message || e.code || JSON.stringify(e)) : String(e);\n const code = (e && typeof e === 'object' && e.code) ? e.code : null;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h('div', {className:'w-9 h-9 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#21734618',color:GREEN}}, h(Icon,{name:'table_chart',size:20})),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-error flex items-center gap-1.5'}, h(Icon,{name:'error',size:14}), 'Excel error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1 break-words'}, msg),\n code && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-1'}, code)\n )\n );\n }\n if (endpoint === 'range') {\n const rows = (data && Array.isArray(data.values)) ? data.values : [];\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No data in used range.');\n const header = rows[0] || [];\n const bodyRows = rows.slice(1);\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[12px] font-bold flex items-center gap-2'},\n h('div', {className:'w-5 h-5 rounded flex items-center justify-center flex-shrink-0', style:{background:GREEN+'22',color:GREEN}}, h(Icon,{name:'grid_on',size:13})),\n data && data.address ? data.address : 'Used range'\n ),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, rows.length + ' rows - ' + header.length + ' cols')\n ),\n h('div', {className:'overflow-x-auto max-h-[480px]'},\n h('table', {className:'w-full text-[12px] border-collapse'},\n h('thead', null,\n h('tr', {className:'sticky top-0'},\n h('th', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-b border-r border-border-subtle bg-surface-container-lowest', style:{width:40}}, ''),\n header.map((c, j) => h('th', {key:j, className:'px-3 py-1.5 text-left font-bold border-b border-border-subtle text-on-surface', style:{background:GREEN+'14'}}, String(c == null ? '' : c)))\n )\n ),\n h('tbody', null,\n bodyRows.slice(0, 200).map((row, i) => h('tr', {key:i, className:'border-b border-border-subtle/50 hover:bg-white/[0.02]'},\n h('td', {className:'px-2 py-1.5 text-text-muted font-mono-data text-right border-r border-border-subtle/50', style:{width:40}}, i+2),\n (Array.isArray(row) ? row : []).map((cell, j) => h('td', {key:j, className:'px-3 py-1.5 whitespace-nowrap'}, String(cell == null ? '' : cell)))\n ))\n )\n )\n )\n );\n }\n if (endpoint === 'worksheets') {\n const sheets = (data && Array.isArray(data.value)) ? data.value : [];\n if (!sheets.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No worksheets.');\n const ordered = sheets.slice().sort((a, b) => (a.position || 0) - (b.position || 0));\n return h('div', {className:'glass-panel rounded-xl p-md mt-3'},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-text-muted mb-3 flex items-center gap-1.5'},\n h(Icon,{name:'tab',size:12}), sheets.length + ' Worksheet' + (sheets.length === 1 ? '' : 's')\n ),\n h('div', {className:'flex flex-wrap gap-2'},\n ordered.map((s, i) => {\n const hidden = s.visibility && s.visibility !== 'Visible';\n return h('div', {\n key: s.id || i,\n className:'flex items-center gap-2 px-3 py-2 rounded-lg border bg-surface-container-lowest ' + (hidden ? 'border-border-subtle/60 opacity-60' : 'border-border-subtle'),\n style: hidden ? {} : {borderColor: GREEN + '55'}\n },\n h('div', {className:'w-2 h-2 rounded-sm flex-shrink-0', style:{background: hidden ? '#888' : GREEN}}),\n h('div', {className:'flex flex-col min-w-0'},\n h('span', {className:'text-[13px] font-bold text-on-surface truncate'}, s.name || '(unnamed)'),\n h('span', {className:'text-[9px] text-text-muted font-mono-data'},\n 'pos ' + (s.position != null ? s.position : '-') + (hidden ? ' - ' + s.visibility : '')\n )\n )\n );\n })\n )\n );\n }\n // Default: workbooks\n const files = (data && Array.isArray(data.value)) ? data.value : [];\n if (!files.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No Excel workbooks found.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(240px, 1fr))'}},\n files.map(f => h('a', {\n key: f.id, href: f.webUrl, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-secondary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#21734618',color:'#217346'}},\n h(Icon,{name:'table_chart',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, f.name || '(unnamed)'),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted flex-wrap'},\n f.id && h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), String(f.id).slice(0, 8) + '-'\n ),\n h('span', {className:'font-mono-data'}, fmtRelative(f.lastModifiedDateTime)),\n f.size != null && h('span', {className:'font-mono-data text-on-surface-variant'}, fmtFileSize(f.size))\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,656],"d2e5ab22-5b4e-11f1-8887-ee650bb23af1","drive.ts","d2e5abd6-5b4e-11f1-8887-ee650bb23af1",524,["D",1780052952897],"// DriveRenderer — per-integration renderer for the 'drive' integration.\nexport const RENDERER = `\nexport function DriveRenderer({data, endpoint}) {\n if (endpoint === 'about' && data?.user) {\n const q = data.storageQuota || {};\n const usedGB = (parseInt(q.usage||0,10) / 1073741824).toFixed(2);\n const limitGB = q.limit ? (parseInt(q.limit,10) / 1073741824).toFixed(0) : '-';\n const pct = q.limit ? Math.min(100, (parseInt(q.usage||0,10) / parseInt(q.limit,10)) * 100) : 0;\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-4 mb-4'},\n data.user.photoLink ? h('img',{src:data.user.photoLink, className:'w-14 h-14 rounded-full border border-border-subtle'}) : null,\n h('div', null,\n h('h3', {className:'font-bold text-[18px]'}, data.user.displayName),\n h('div', {className:'text-[12px] text-text-muted'}, data.user.emailAddress)\n )\n ),\n h('div', {className:'space-y-2'},\n h('div', {className:'flex justify-between text-[12px]'},\n h('span', {className:'text-on-surface-variant'}, 'Storage'),\n h('span', {className:'font-mono-data font-bold'}, usedGB + ' GB / ' + limitGB + ' GB')\n ),\n h('div', {className:'h-2 bg-surface-container-high rounded-full overflow-hidden'},\n h('div', {className:'h-full transition-all', style:{width: pct + '%', background: pct > 90 ? '#ffb4ab' : '#4edea3'}})\n )\n )\n );\n }\n const files = data?.files || [];\n if (!files.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No files.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(220px, 1fr))'}},\n files.map(f => {\n const ico = driveIconFor(f.mimeType);\n const owner = f.owners?.[0];\n return h('a', {\n key: f.id, href: f.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex flex-col gap-2 hover:border-primary/30 transition-all group'\n },\n h('div', {className:'flex items-start justify-between gap-2'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background: ico.color+'18', color: ico.color}},\n h(Icon,{name: ico.icon, size:22})\n ),\n h(Icon, {name:'open_in_new', size:14, className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0 mt-1'})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold text-on-surface truncate'}, f.name),\n h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, f.mimeType?.replace('application/vnd.google-apps.','').replace('application/','') || '-')\n ),\n h('div', {className:'flex items-center justify-between text-[10px] text-text-muted pt-2 border-t border-border-subtle'},\n h('span', {className:'truncate flex items-center gap-1'},\n owner?.photoLink ? h('img',{src:owner.photoLink, className:'w-4 h-4 rounded-full'}) : null,\n owner?.displayName || '-'\n ),\n h('span', {className:'font-mono-data flex-shrink-0'}, fmtRelative(f.modifiedTime)),\n f.size && h('span', {className:'font-mono-data text-on-surface-variant flex-shrink-0'}, fmtFileSize(f.size))\n ),\n (f.shared || f.starred) && h('div', {className:'flex gap-1 flex-wrap'},\n f.starred && h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-tertiary/10 text-tertiary'}, h(Icon,{name:'star',size:9,filled:true}), 'STARRED'),\n f.shared && h('span', {className:'inline-flex items-center gap-0.5 text-[9px] px-1.5 py-0.5 rounded bg-secondary/10 text-secondary'}, h(Icon,{name:'group',size:9}), 'SHARED')\n )\n );\n })\n )\n );\n}\n`;\n",[],[359,664],"24f6f7ae-5b54-11f1-a375-ee650bb23af1","docs.ts","d2841768-5b4e-11f1-a491-ee650bb23af1",549,["D",1780055238021],"// DocsRenderer — per-integration renderer for the 'docs' integration.\nexport const RENDERER = `\nexport function DocsRenderer({data, endpoint}) {\n if (endpoint === 'document' && data?.body) {\n const blocks = data.body.content || [];\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3'},\n h('div', {className:'flex items-center gap-3 mb-4 pb-3 border-b border-border-subtle'},\n h('div', {className:'w-10 h-10 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#4285f418',color:'#4285f4'}},\n h(Icon,{name:'description',size:22})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('h3', {className:'text-[16px] font-bold truncate'}, data.title),\n h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, blocks.length + ' content blocks - revision ' + (data.revisionId?.slice(0,8) || '-'))\n )\n ),\n h('div', {className:'space-y-2 max-h-[480px] overflow-y-auto text-[13px] text-on-surface-variant leading-relaxed'},\n blocks.slice(0,30).map((b,i) => {\n const text = (b.paragraph?.elements || []).map(e => e.textRun?.content || '').join('');\n if (!text.trim()) return null;\n const isH1 = b.paragraph?.paragraphStyle?.namedStyleType === 'HEADING_1';\n const isH2 = b.paragraph?.paragraphStyle?.namedStyleType === 'HEADING_2';\n return h('p', {key:i, className: isH1?'text-[18px] font-bold text-on-surface':isH2?'text-[15px] font-bold text-on-surface mt-2':'text-[13px]'}, text);\n })\n )\n );\n }\n const docs = data?.files || [];\n if (!docs.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No documents.');\n return h('div', {className:'mt-3'},\n h('div', {className:'grid gap-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(260px, 1fr))'}},\n docs.map(d => h('a', {\n key:d.id, href:d.webViewLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-xl p-md flex items-start gap-3 hover:border-primary/40 transition-all group'\n },\n h('div', {className:'w-11 h-11 rounded-lg flex items-center justify-center flex-shrink-0', style:{background:'#4285f418',color:'#4285f4'}},\n h(Icon,{name:'description',size:24})\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, d.name),\n h('div', {className:'flex items-center gap-2 mt-1 text-[10px] text-text-muted'},\n h('span', {className:'font-mono-data flex items-center gap-1'},\n h(Icon,{name:'fingerprint',size:10}), (d.id||'').slice(0,8) + '-'\n ),\n h('span', null, '-'),\n h('span', null, fmtRelative(d.modifiedTime))\n ),\n d.owners?.[0] && h('div', {className:'text-[10px] text-on-surface-variant truncate mt-1 flex items-center gap-1'},\n h(Icon,{name:'person',size:11}), d.owners[0].displayName\n )\n ),\n h(Icon,{name:'open_in_new',size:14,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ))\n )\n );\n}\n`;\n",[],[359,672],"d224620a-5b4e-11f1-95e0-ee650bb23af1","calendar.ts","d22462dc-5b4e-11f1-95e0-ee650bb23af1",522,["D",1780052951618],"// CalendarRenderer — per-integration renderer for the 'calendar' integration.\nexport const RENDERER = `\nexport function CalendarRenderer({data, endpoint}) {\n if (endpoint === 'calendars' && Array.isArray(data?.items)) {\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 space-y-2'},\n h('h3', {className:'font-bold text-[15px] mb-2'}, data.items.length + ' Calendars'),\n data.items.map(c => h('div', {key:c.id, className:'flex items-center gap-3 p-2.5 rounded-lg bg-surface-container/40'},\n h('div', {className:'w-3 h-3 rounded flex-shrink-0', style:{background: c.backgroundColor || '#c0c1ff'}}),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] font-bold truncate'}, c.summary, c.primary && h('span', {className:'ml-2 text-[9px] text-primary uppercase'}, 'PRIMARY')),\n h('div', {className:'text-[11px] text-text-muted truncate'}, c.id)\n ),\n h('span', {className:'text-[10px] text-on-surface-variant uppercase'}, c.accessRole)\n ))\n );\n }\n const events = data?.items || [];\n if (!events.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No events scheduled.');\n const groups = {};\n events.forEach(e => {\n const startISO = e.start?.dateTime || e.start?.date;\n const dKey = startISO ? new Date(startISO).toDateString() : 'undated';\n if (!groups[dKey]) groups[dKey] = [];\n groups[dKey].push(e);\n });\n const todayStr = new Date().toDateString();\n const tomStr = new Date(Date.now()+86400000).toDateString();\n const dateLabel = (d) => d === todayStr ? 'TODAY' : d === tomStr ? 'TOMORROW' : new Date(d).toLocaleDateString('en-US',{weekday:'long',month:'short',day:'numeric'}).toUpperCase();\n return h('div', {className:'mt-3 space-y-4'},\n Object.entries(groups).map(([dKey, evs]) => h('div', {key:dKey},\n h('div', {className:'text-[10px] font-bold uppercase tracking-widest text-secondary mb-2 sticky top-0 bg-background/80 backdrop-blur-sm py-1'},\n dateLabel(dKey), h('span', {className:'ml-2 text-text-muted'}, '- ' + evs.length + ' event' + (evs.length===1?'':'s'))\n ),\n h('div', {className:'space-y-2'},\n evs.map(e => {\n const startT = e.start?.dateTime ? new Date(e.start.dateTime).toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : 'All-day';\n const endT = e.end?.dateTime ? new Date(e.end.dateTime).toLocaleTimeString('en-US',{hour:'numeric',minute:'2-digit'}) : '';\n const attendees = (e.attendees||[]).filter(a => !a.self);\n return h('a', {\n key:e.id, href:e.htmlLink, target:'_blank', rel:'noopener',\n className:'glass-panel rounded-lg p-3 flex items-start gap-3 hover:border-secondary/30 transition-all group'\n },\n h('div', {className:'flex flex-col items-center justify-center w-16 flex-shrink-0 pr-2 border-r border-border-subtle'},\n h('div', {className:'text-[12px] font-bold text-on-surface font-mono-data'}, startT),\n endT && h('div', {className:'text-[10px] text-text-muted font-mono-data mt-0.5'}, '- ' + endT)\n ),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[14px] font-bold text-on-surface truncate'}, e.summary || '(no title)'),\n e.location && h('div', {className:'text-[11px] text-text-muted truncate mt-0.5 flex items-center gap-1'},\n h(Icon,{name:'place',size:11}), e.location\n ),\n e.description && h('div', {className:'text-[11px] text-on-surface-variant truncate mt-0.5'}, e.description.replace(/<[^>]+>/g,'').slice(0,80)),\n attendees.length > 0 && h('div', {className:'flex items-center gap-1 mt-1.5 text-[10px] text-text-muted'},\n h(Icon,{name:'group',size:11}),\n attendees.length + ' attendee' + (attendees.length===1?'':'s'),\n e.hangoutLink && h('span', {className:'ml-1 text-secondary'}, '- Meet')\n )\n ),\n e.status === 'confirmed' && h('div', {className:'w-1.5 h-1.5 rounded-full bg-secondary mt-2 flex-shrink-0'})\n );\n })\n )\n ))\n );\n}\n`;\n",[],[359,680],"d19debbc-5b4e-11f1-a8fd-ee650bb23af1","ahrefs.ts","d19ded24-5b4e-11f1-a8fd-ee650bb23af1",521,["D",1780052950747],"// AhrefsRenderer — per-integration renderer for the 'ahrefs' integration.\nexport const RENDERER = `\nexport function AhrefsRenderer({data, endpoint}) {\n const AH = '#ff6924';\n if (!data) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No data.');\n if (data.error) return h('div', {className:'glass-panel rounded-xl p-lg mt-3 border border-error/40 flex items-start gap-3'},\n h(Icon,{name:'error',size:18,className:'text-error flex-shrink-0 mt-0.5'}),\n h('div', null,\n h('div', {className:'text-[13px] font-bold text-error'}, 'Ahrefs API error'),\n h('div', {className:'text-[12px] text-on-surface-variant mt-1'}, String(data.error))\n )\n );\n\n const num = (n) => (n == null || n === '') ? '-' : Number(n).toLocaleString();\n const compact = (n) => {\n const v = Number(n) || 0;\n if (v >= 1e6) return (v/1e6).toFixed(1).replace(/\\\\.0$/,'') + 'M';\n if (v >= 1e3) return (v/1e3).toFixed(1).replace(/\\\\.0$/,'') + 'K';\n return String(v);\n };\n const drColor = (dr) => dr >= 70 ? '#4edea3' : dr >= 40 ? '#fbbc04' : dr >= 20 ? '#ffa726' : '#94a3b8';\n const hostOf = (u) => { try { return new URL(u).hostname.replace(/^www\\\\./,''); } catch { return u || ''; } };\n\n const chip = (label, value, icon) => h('div', {key:label, className:'glass-panel rounded-xl p-md flex flex-col gap-1 border border-border-subtle'},\n h('div', {className:'flex items-center gap-1.5 text-[10px] uppercase tracking-widest text-text-muted'},\n icon && h(Icon,{name:icon,size:12,style:{color:AH}}), label\n ),\n h('div', {className:'text-[22px] font-bold text-on-surface font-mono-data'}, value)\n );\n const chipGrid = (chips) => h('div', {className:'grid gap-3 mt-3', style:{gridTemplateColumns:'repeat(auto-fill, minmax(150px, 1fr))'}}, chips.filter(Boolean));\n\n // -- domain-rating: gauge + rank ------------------------------------------\n if (endpoint === 'domain-rating') {\n const dr = data.domain_rating || data;\n const score = dr.domain_rating != null ? Number(dr.domain_rating) : null;\n const c = drColor(score || 0);\n return h('div', {className:'glass-panel rounded-xl p-lg mt-3 flex items-center gap-6'},\n h('div', {className:'relative w-24 h-24 rounded-full flex items-center justify-center flex-shrink-0',\n style:{background:'conic-gradient('+c+' '+(score||0)*3.6+'deg, rgba(255,255,255,0.07) 0deg)'}},\n h('div', {className:'w-[78px] h-[78px] rounded-full bg-surface-container flex flex-col items-center justify-center'},\n h('span', {className:'text-[28px] font-bold font-mono-data', style:{color:c}}, score != null ? score : '-'),\n h('span', {className:'text-[9px] uppercase tracking-widest text-text-muted'}, 'DR')\n )\n ),\n h('div', {className:'flex-1'},\n h('h3', {className:'text-[16px] font-bold'}, 'Domain Rating'),\n h('div', {className:'text-[12px] text-text-muted mt-0.5'}, 'Backlink strength on a 0-100 scale'),\n dr.ahrefs_rank != null && h('div', {className:'mt-3 inline-flex items-center gap-1.5 px-3 py-1.5 rounded-lg bg-surface-container-high text-[12px] font-mono-data'},\n h(Icon,{name:'trending_up',size:13,style:{color:AH}}), 'Ahrefs Rank #' + num(dr.ahrefs_rank)\n )\n )\n );\n }\n\n // -- site-explorer / backlinks-stats: metric chips ------------------------\n if (endpoint === 'site-explorer' || endpoint === 'backlinks-stats') {\n const m = data.metrics || data;\n const chips = endpoint === 'site-explorer' ? [\n m.org_traffic != null && chip('Org Traffic', compact(m.org_traffic), 'visibility'),\n m.org_keywords != null && chip('Org Keywords', num(m.org_keywords), 'key'),\n m.org_cost != null && chip('Traffic Value', '$' + compact(m.org_cost), 'paid'),\n m.paid_traffic != null && chip('Paid Traffic', compact(m.paid_traffic), 'ads_click'),\n m.paid_keywords != null && chip('Paid Keywords', num(m.paid_keywords), 'campaign'),\n m.paid_cost != null && chip('Paid Cost', '$' + compact(m.paid_cost), 'attach_money')\n ] : [\n m.live != null && chip('Live Backlinks', compact(m.live), 'link'),\n m.all_time != null && chip('All-Time', compact(m.all_time), 'history'),\n m.live_refdomains != null && chip('Live RefDomains', compact(m.live_refdomains), 'public'),\n m.all_time_refdomains != null && chip('All-Time RefDomains', compact(m.all_time_refdomains), 'language')\n ];\n if (!chips.filter(Boolean).length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No metrics returned.');\n return chipGrid(chips);\n }\n\n // -- ranked list builder --------------------------------------------------\n const rankedRows = (rows, opts) => {\n const max = Math.max(1, ...rows.map(opts.metric));\n return h('div', {className:'glass-panel rounded-xl mt-3 overflow-hidden'},\n h('div', {className:'p-3 border-b border-border-subtle bg-surface-container-lowest/50 flex justify-between items-center'},\n h('span', {className:'text-[13px] font-bold flex items-center gap-2'}, h(Icon,{name:opts.icon,size:16,style:{color:AH}}), opts.title),\n h('span', {className:'text-[10px] text-text-muted font-mono-data'}, rows.length + ' ' + opts.unit)\n ),\n h('div', {className:'divide-y divide-border-subtle/50 max-h-[560px] overflow-y-auto'},\n rows.map((r, i) => {\n const pct = Math.max(2, (opts.metric(r) / max) * 100);\n const lbl = opts.label(r);\n return h(opts.href ? 'a' : 'div', Object.assign({key:i,\n className:'block px-3.5 py-2.5 hover:bg-white/[0.02] transition-colors ' + (opts.href ? 'group' : '')},\n opts.href ? {href:opts.href(r), target:'_blank', rel:'noreferrer'} : {}),\n h('div', {className:'flex items-center gap-3'},\n h('span', {className:'text-[10px] text-text-muted font-mono-data w-5 text-right flex-shrink-0'}, i+1),\n opts.badge && h('span', {className:'text-[10px] font-bold font-mono-data px-1.5 py-0.5 rounded flex-shrink-0', style:opts.badge(r)}, opts.badgeText(r)),\n h('div', {className:'flex-1 min-w-0'},\n h('div', {className:'text-[13px] truncate text-on-surface ' + (opts.href?'group-hover:text-primary transition-colors':'')}, lbl),\n opts.sub && h('div', {className:'text-[10px] text-text-muted truncate mt-0.5'}, opts.sub(r))\n ),\n h('div', {className:'text-right flex-shrink-0'},\n h('div', {className:'text-[12px] font-bold font-mono-data text-on-surface'}, opts.metricText(r)),\n opts.metricLabel && h('div', {className:'text-[9px] text-text-muted uppercase'}, opts.metricLabel)\n ),\n opts.href && h(Icon,{name:'open_in_new',size:12,className:'text-text-muted opacity-0 group-hover:opacity-100 transition-opacity flex-shrink-0'})\n ),\n h('div', {className:'h-1 mt-1.5 rounded-full bg-surface-container-high overflow-hidden ml-8'},\n h('div', {className:'h-full rounded-full', style:{width:pct+'%', background:opts.barColor ? opts.barColor(r) : AH}})\n )\n );\n })\n )\n );\n };\n\n // -- keywords / organic-keywords ------------------------------------------\n if (endpoint === 'keywords' || endpoint === 'organic-keywords') {\n const rows = data.keywords || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No keywords found.');\n const metric = (r) => Number(r.sum_traffic ?? r.traffic_potential ?? r.volume ?? r.global_volume ?? 0);\n return rankedRows(rows, {\n title:'Keywords', icon:'key', unit:'keywords',\n metric, metricText:(r)=>compact(metric(r)), metricLabel: (rows[0].sum_traffic!=null||rows[0].traffic_potential!=null)?'traffic':'volume',\n label:(r)=>r.keyword || '-',\n sub:(r)=>{\n const parts = [];\n if (r.volume != null) parts.push('Vol ' + compact(r.volume));\n const diff = r.keyword_difficulty ?? r.difficulty;\n if (diff != null) parts.push('KD ' + diff);\n if (r.cpc != null) parts.push('$' + Number(r.cpc).toFixed(2) + ' CPC');\n if (r.best_position != null) parts.push('Pos #' + r.best_position);\n return parts.join(' - ');\n },\n badge:(r)=>{ const d = Number(r.keyword_difficulty ?? r.difficulty ?? 0); const c = drColor(100-d); return {background:c+'22',color:c}; },\n badgeText:(r)=>'KD ' + (r.keyword_difficulty ?? r.difficulty ?? '-')\n });\n }\n\n // -- top-pages ------------------------------------------------------------\n if (endpoint === 'top-pages') {\n const rows = data.pages || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No pages found.');\n const metric = (r) => Number(r.sum_traffic ?? r.value ?? 0);\n return rankedRows(rows, {\n title:'Top Pages', icon:'article', unit:'pages',\n metric, metricText:(r)=>compact(metric(r)), metricLabel:'traffic',\n href:(r)=>r.url, label:(r)=>hostOf(r.url) + (function(){ try { return new URL(r.url).pathname; } catch { return ''; } })(),\n sub:(r)=>{\n const parts = [];\n if (r.top_keyword) parts.push('Top: ' + r.top_keyword + (r.top_keyword_best_position!=null?' (#'+r.top_keyword_best_position+')':''));\n if (r.keywords != null) parts.push(compact(r.keywords) + ' keywords');\n return parts.join(' - ');\n }\n });\n }\n\n // -- backlinks / referring-domains ----------------------------------------\n if (endpoint === 'backlinks' || endpoint === 'referring-domains') {\n const rows = data.backlinks || data.refdomains || (Array.isArray(data) ? data : []);\n if (!rows.length) return h('div', {className:'text-center text-text-muted text-[13px] py-8 mt-3'}, 'No backlinks found.');\n const isRef = endpoint === 'referring-domains';\n const dr = (r) => Number(r.domain_rating ?? r.domain_rating_source ?? 0);\n return rankedRows(rows, {\n title: isRef ? 'Referring Domains' : 'Backlinks', icon: isRef ? 'public' : 'link', unit: isRef ? 'domains' : 'backlinks',\n metric: dr, metricText:(r)=>dr(r) || '-', metricLabel:'DR', barColor:(r)=>drColor(dr(r)),\n href:(r)=> isRef ? ('https://' + (r.domain||'')) : r.url_from,\n label:(r)=> isRef ? (r.domain || '-') : (r.title || r.name_source || hostOf(r.url_from)),\n sub:(r)=>{\n if (isRef) {\n const parts = [];\n if (r.dofollow_links != null) parts.push(compact(r.dofollow_links) + ' dofollow links');\n if (r.traffic_domain != null) parts.push(compact(r.traffic_domain) + ' traffic');\n return parts.join(' - ');\n }\n const parts = [];\n if (r.anchor) parts.push('\"' + r.anchor.slice(0,50) + '\"');\n parts.push(hostOf(r.url_from));\n return parts.join(' - ');\n },\n badge:(r)=> (!isRef && r.is_dofollow != null) ? {background:(r.is_dofollow?'#4edea3':'#94a3b8')+'22', color:(r.is_dofollow?'#4edea3':'#94a3b8')} : null,\n badgeText:(r)=> r.is_dofollow ? 'FOLLOW' : 'NOFOLLOW'\n });\n }\n\n // -- fallback -------------------------------------------------------------\n return h('div', {className:'glass-panel rounded-xl p-md mt-3 text-[12px] text-text-muted font-mono-data overflow-x-auto'},\n h('pre', null, JSON.stringify(data, null, 2).slice(0, 1200))\n );\n}\n`;\n",[],[359,688],"d0f3f4b8-5b4e-11f1-8d86-ee650bb23af1","_preamble.ts","d0f3f576-5b4e-11f1-8d86-ee650bb23af1",["D",1780052949625],"// Shared renderer preamble: React import, h, Icon, normalizeData, format helpers.\n// Consumed by frontend_renderers.ts + frontend_renderers2.ts (no build step on val.town).\nexport const PREAMBLE = `\n\nimport React from 'react';\nconst h = React.createElement;\nconst _ui = await import('/api/app-ui');\nconst { Icon } = _ui;\n\n// --- DATA NORMALIZATION -----------------------------------------------------\n// For integrations without their own custom renderer, flatten nested API responses\n// into rows the generic DataDisplay can render as a table.\nexport function normalizeData(name, endpoint, data) {\n if (!data || typeof data !== 'object') return data;\n switch (name) {\n case 'gsc':\n if (Array.isArray(data.rows)) return data.rows.map(r => ({\n [endpoint === 'pages' ? 'page' : 'query']: (r.keys || [])[0] || '',\n clicks: r.clicks, impressions: r.impressions,\n ctr: r.ctr != null ? (r.ctr * 100).toFixed(2) + '%' : '-',\n position: r.position != null ? r.position.toFixed(1) : '-'\n }));\n if (Array.isArray(data.sitemap)) return data.sitemap.map(s => ({\n path: s.path, type: s.type, lastSubmitted: s.lastSubmitted?.slice(0,10),\n isPending: s.isPending, warnings: s.warnings, errors: s.errors, contents: s.contents?.length\n }));\n return data;\n case 'youtube':\n if (Array.isArray(data.items)) return data.items.map(item => ({\n id: item.id?.videoId || item.id?.channelId || (typeof item.id === 'string' ? item.id : ''),\n title: item.snippet?.title || '', channel: item.snippet?.channelTitle || '',\n published: item.snippet?.publishedAt?.slice(0,10) || '',\n views: item.statistics?.viewCount || '', likes: item.statistics?.likeCount || '',\n comments: item.statistics?.commentCount || '', subscribers: item.statistics?.subscriberCount || ''\n }));\n return data;\n case 'producthunt': {\n const edges = (data?.data?.posts?.edges?.length ? data.data.posts.edges : data?.data?.topics?.edges) || [];\n if (edges.length) return edges.map(e => {\n const n = e.node || e;\n return { id:n.id, name:n.name, tagline:n.tagline||'', slug:n.slug||'', votes:n.votesCount||'', followers:n.followersCount||'', comments:n.commentsCount||'', url:n.url||'' };\n });\n return data;\n }\n case 'wordpress':\n if (Array.isArray(data)) return data.map(p => ({\n id:p.id, date:p.date?.slice(0,10)||'', status:p.status||'',\n title:p.title?.rendered||p.title||'', link:p.link||'', type:p.type||'',\n author:p.author, commentCount:p.comment_count\n }));\n return data;\n case 'outlook':\n if (Array.isArray(data.value)) return data.value.map(m => ({\n subject: m.subject || '(no subject)',\n from: m.from?.emailAddress?.name || m.from?.emailAddress?.address || '',\n received: m.receivedDateTime?.slice(0,16) || m.start?.dateTime?.slice(0,16) || '',\n isRead: m.isRead != null ? (m.isRead ? 'Yes' : 'No') : '',\n location: m.location?.displayName || '',\n end: m.end?.dateTime?.slice(0,16) || ''\n }));\n return data;\n case 'linkedin':\n if (Array.isArray(data.elements)) return data.elements.map(e => ({\n id: e.id || '', urn: e.ugcPost || e.author || '',\n created: e.created?.time ? new Date(e.created.time).toISOString().slice(0,10) : '',\n lifecycle: e.lifecycleState || ''\n }));\n if (data.localizedFirstName || data.firstName) {\n const first = data.localizedFirstName || Object.values(data.firstName?.localized || {})[0] || '';\n const last = data.localizedLastName || Object.values(data.lastName?.localized || {})[0] || '';\n return { name: (first + ' ' + last).trim(), id: data.id, headline: data.localizedHeadline || '' };\n }\n return data;\n default: return data;\n }\n}\n\n// --- HELPERS -----------------------------------------------------------------\nconst fmtRelative = (iso) => {\n if (!iso) return '';\n const ms = Date.now() - new Date(iso).getTime();\n if (ms < 60000) return 'just now';\n if (ms < 3600000) return Math.floor(ms/60000) + 'm ago';\n if (ms < 86400000) return Math.floor(ms/3600000) + 'h ago';\n if (ms < 604800000) return Math.floor(ms/86400000) + 'd ago';\n return new Date(iso).toLocaleDateString('en-US', {month:'short', day:'numeric'});\n};\nconst fmtFileSize = (bytes) => {\n const n = parseInt(bytes,10);\n if (!n) return '-';\n if (n < 1024) return n + ' B';\n if (n < 1048576) return (n/1024).toFixed(1) + ' KB';\n if (n < 1073741824) return (n/1048576).toFixed(1) + ' MB';\n return (n/1073741824).toFixed(2) + ' GB';\n};\nconst initialsOf = (name) => (name||'?').replace(/<[^>]+>/g,'').trim().split(/\\\\s+/).slice(0,2).map(s=>s[0]||'').join('').toUpperCase() || '?';\nconst colorFromName = (name) => {\n const colors = ['#c0c1ff','#4edea3','#d0bcff','#8083ff','#6ffbbe','#a078ff','#e1e0ff','#494bd6'];\n const sum = (name||'').split('').reduce((a,c) => a + c.charCodeAt(0), 0);\n return colors[sum % colors.length];\n};\n`;\n",[],[359,696],["D",1780052949625],[425,426,427,428,429,430,431,432,433,434,435,436,437,438,439,440,441],[359],"5d56fa0c-5a6b-11f1-bd26-ee650bb23af1","tasks.ts","97a41542-589e-11f1-b2d9-ee650bb23af1",434,["D",1779955259903],"// Tasks — Phase 1 Framework Layer.\n// Lightweight task tracking that can inherit ICE scoring from a parent\n// experiment, and that the LLM can populate from a KR gap or experiment\n// hypothesis so users don't have to draft work themselves.\nimport { sqlite } from \"../db.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\nimport { propagateTaskDone } from \"./propagate.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport interface Task {\n id: string;\n title: string;\n description: string | null;\n status: string; // todo | doing | done | blocked\n priority: string; // low | medium | high | urgent\n workspace: string | null;\n assignee: string | null;\n due_date: string | null;\n objective_id: string | null;\n key_result_id: string | null;\n experiment_id: string | null;\n ice_score: number | null;\n impact: number | null;\n confidence: number | null;\n ease: number | null;\n tags: string[];\n created_at: string;\n updated_at: string;\n completed_at: string | null;\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp10 = (n: any) => {\n const v = Number(n);\n if (!Number.isFinite(v)) return null;\n return Math.max(1, Math.min(10, Math.round(v)));\n};\n\nfunction parseTags(raw: any): string[] {\n if (!raw) return [];\n if (Array.isArray(raw)) return raw.map(String);\n try {\n const parsed = JSON.parse(String(raw));\n return Array.isArray(parsed) ? parsed.map(String) : [];\n } catch {\n return [];\n }\n}\n\nfunction rowToTask(columns: string[], row: any[]): Task {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n status: String(obj.status ?? \"todo\"),\n priority: String(obj.priority ?? \"medium\"),\n workspace: obj.workspace ?? null,\n assignee: obj.assignee ?? null,\n due_date: obj.due_date ?? null,\n objective_id: obj.objective_id ?? null,\n key_result_id: obj.key_result_id ?? null,\n experiment_id: obj.experiment_id ?? null,\n ice_score: obj.ice_score != null ? Number(obj.ice_score) : null,\n impact: obj.impact != null ? Number(obj.impact) : null,\n confidence: obj.confidence != null ? Number(obj.confidence) : null,\n ease: obj.ease != null ? Number(obj.ease) : null,\n tags: parseTags(obj.tags),\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n completed_at: obj.completed_at ?? null,\n };\n}\n\nconst TASK_COLS = `id, title, description, status, priority, workspace, assignee,\n due_date, objective_id, key_result_id, experiment_id, ice_score,\n impact, confidence, ease, tags, created_at, updated_at, completed_at`;\n\nasync function getTaskById(id: string): Promise<Task | null> {\n const r = await sqlite.execute(\n `SELECT ${TASK_COLS} FROM tasks WHERE id = ?`,\n [id],\n );\n if (!r.rows?.length) return null;\n return rowToTask(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listTasks(filters: {\n workspace?: string;\n status?: string;\n assignee?: string;\n objective_id?: string;\n experiment_id?: string;\n} = {}): Promise<Task[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.workspace) { where.push(\"workspace = ?\"); args.push(filters.workspace); }\n if (filters.status) { where.push(\"status = ?\"); args.push(filters.status); }\n if (filters.assignee) { where.push(\"assignee = ?\"); args.push(filters.assignee); }\n if (filters.objective_id) { where.push(\"objective_id = ?\"); args.push(filters.objective_id); }\n if (filters.experiment_id){ where.push(\"experiment_id = ?\");args.push(filters.experiment_id); }\n // ORDER BY: due_date ASC NULLS LAST, then created_at DESC.\n // SQLite sorts NULLs first by default, so we use a CASE expression.\n const sql = `SELECT ${TASK_COLS}\n FROM tasks\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY (CASE WHEN due_date IS NULL THEN 1 ELSE 0 END) ASC,\n due_date ASC,\n created_at DESC`;\n const r = await sqlite.execute(sql, args);\n return (r.rows || []).map((row) => rowToTask(r.columns, row as any[]));\n } catch {\n return [];\n }\n}\n\nexport async function listIceRankedTasks(limit = 20): Promise<Task[]> {\n try {\n const r = await sqlite.execute(\n `SELECT ${TASK_COLS}\n FROM tasks\n WHERE ice_score IS NOT NULL\n ORDER BY ice_score DESC\n LIMIT ?`,\n [limit],\n );\n return (r.rows || []).map((row) => rowToTask(r.columns, row as any[]));\n } catch {\n return [];\n }\n}\n\n// ─── Write ───────────────────────────────────────────────────────────────────\n\nexport async function createTask(data: Partial<Task>): Promise<Task> {\n // NOTE: tasks.id is INTEGER AUTOINCREMENT (db.ts) — do NOT supply a UUID here\n // (that caused SQLITE_MISMATCH and broke all task creation). Let SQLite assign it.\n const now = new Date().toISOString();\n\n let impact = data.impact ?? null;\n let confidence = data.confidence ?? null;\n let ease = data.ease ?? null;\n let ice_score = data.ice_score ?? null;\n\n // Inherit ICE from the parent experiment when one is linked + caller didn't\n // override. Lets users say \"make a task for this experiment\" without losing\n // the experiment's prioritization context.\n if (data.experiment_id && (impact == null || confidence == null || ease == null || ice_score == null)) {\n try {\n const exp = await sqlite.execute(\n \"SELECT impact, confidence, ease FROM experiments WHERE id = ?\",\n [data.experiment_id],\n );\n if (exp.rows?.length) {\n const row = exp.rows[0] as any[];\n const expI = Number(row[0]); const expC = Number(row[1]); const expE = Number(row[2]);\n if (impact == null) impact = expI;\n if (confidence == null) confidence = expC;\n if (ease == null) ease = expE;\n if (ice_score == null && Number.isFinite(expI) && Number.isFinite(expC) && Number.isFinite(expE)) {\n ice_score = Math.round(((expI + expC + expE) / 3) * 100) / 100;\n }\n }\n } catch { /* non-critical */ }\n }\n\n // If impact/confidence/ease are all set but ice_score isn't, compute it.\n if (ice_score == null && impact != null && confidence != null && ease != null) {\n const i = clamp10(impact); const c = clamp10(confidence); const e = clamp10(ease);\n if (i !== null && c !== null && e !== null) {\n ice_score = Math.round(((i + c + e) / 3) * 100) / 100;\n }\n }\n\n const tags = JSON.stringify(Array.isArray(data.tags) ? data.tags : []);\n const status = data.status ?? \"todo\";\n const completed_at = status === \"done\" ? now : null;\n\n await sqlite.execute(\n `INSERT INTO tasks (title, description, status, priority, workspace, assignee,\n due_date, objective_id, key_result_id, experiment_id, ice_score,\n impact, confidence, ease, tags, created_at, updated_at, completed_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled Task\",\n data.description ?? null,\n status,\n data.priority ?? \"medium\",\n data.workspace ?? null,\n data.assignee ?? null,\n data.due_date ?? null,\n data.objective_id ?? null,\n data.key_result_id ?? null,\n data.experiment_id ?? null,\n ice_score,\n impact, confidence, ease,\n tags,\n now, now, completed_at,\n ],\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((idRow.rows[0] as any[])[0]);\n return (await getTaskById(id))!;\n}\n\nexport async function updateTask(\n id: string,\n data: Partial<Task>,\n): Promise<Task | null> {\n try {\n const cur = await getTaskById(id);\n if (!cur) return null;\n const now = new Date().toISOString();\n\n const nextStatus = data.status ?? cur.status;\n // Transition to 'done' stamps completed_at; reverting clears it.\n const completed_at = (() => {\n if (nextStatus === \"done\" && cur.status !== \"done\") return now;\n if (nextStatus !== \"done\" && cur.status === \"done\") return null;\n return cur.completed_at;\n })();\n\n const next = {\n title: data.title ?? cur.title,\n description: \"description\" in data ? (data.description ?? null) : cur.description,\n status: nextStatus,\n priority: data.priority ?? cur.priority,\n workspace: \"workspace\" in data ? (data.workspace ?? null) : cur.workspace,\n assignee: \"assignee\" in data ? (data.assignee ?? null) : cur.assignee,\n due_date: \"due_date\" in data ? (data.due_date ?? null) : cur.due_date,\n objective_id: \"objective_id\" in data ? (data.objective_id ?? null) : cur.objective_id,\n key_result_id: \"key_result_id\" in data ? (data.key_result_id ?? null) : cur.key_result_id,\n experiment_id: \"experiment_id\" in data ? (data.experiment_id ?? null) : cur.experiment_id,\n impact: data.impact != null ? clamp10(data.impact) : cur.impact,\n confidence: data.confidence != null ? clamp10(data.confidence) : cur.confidence,\n ease: data.ease != null ? clamp10(data.ease) : cur.ease,\n tags: \"tags\" in data\n ? JSON.stringify(Array.isArray(data.tags) ? data.tags : [])\n : JSON.stringify(cur.tags),\n };\n\n // Recompute ICE if any of the three components are now present.\n let ice_score = cur.ice_score;\n if (next.impact != null && next.confidence != null && next.ease != null) {\n ice_score = Math.round(((next.impact + next.confidence + next.ease) / 3) * 100) / 100;\n } else if (\"ice_score\" in data) {\n ice_score = data.ice_score ?? null;\n }\n\n await sqlite.execute(\n `UPDATE tasks SET\n title = ?, description = ?, status = ?, priority = ?, workspace = ?,\n assignee = ?, due_date = ?, objective_id = ?, key_result_id = ?,\n experiment_id = ?, ice_score = ?, impact = ?, confidence = ?, ease = ?,\n tags = ?, updated_at = ?, completed_at = ?\n WHERE id = ?`,\n [\n next.title, next.description, next.status, next.priority, next.workspace,\n next.assignee, next.due_date, next.objective_id, next.key_result_id,\n next.experiment_id, ice_score, next.impact, next.confidence, next.ease,\n next.tags, now, completed_at, id,\n ],\n );\n // Closed-loop: a task moving INTO 'done' propagates to the measure layer\n // (executed-work counter + calendar marker + objective progress refresh).\n if (nextStatus === \"done\" && cur.status !== \"done\") {\n await propagateTaskDone({\n title: next.title,\n workspace: next.workspace,\n objective_id: next.objective_id,\n key_result_id: next.key_result_id,\n });\n }\n return await getTaskById(id);\n } catch {\n return null;\n }\n}\n\nexport async function deleteTask(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM tasks WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── LLM-generated task list ────────────────────────────────────────────────\n\ninterface GeneratedTask {\n title: string;\n description?: string;\n priority?: string;\n impact?: number;\n confidence?: number;\n ease?: number;\n}\n\nfunction parseTasksJson(raw: string): GeneratedTask[] {\n const tryParse = (s: string): GeneratedTask[] | null => {\n try {\n const parsed = JSON.parse(s);\n if (Array.isArray(parsed)) return parsed as GeneratedTask[];\n if (parsed && Array.isArray(parsed.tasks)) return parsed.tasks as GeneratedTask[];\n return null;\n } catch { return null; }\n };\n const direct = tryParse(raw);\n if (direct) return direct;\n // Regex-fenced fallback: pull the first JSON array out of mixed text.\n const arrMatch = raw.match(/\\[[\\s\\S]*\\]/);\n if (arrMatch) {\n const parsed = tryParse(arrMatch[0]);\n if (parsed) return parsed;\n }\n const objMatch = raw.match(/\\{[\\s\\S]*\\}/);\n if (objMatch) {\n const parsed = tryParse(objMatch[0]);\n if (parsed) return parsed;\n }\n return [];\n}\n\nexport async function generateTasksFromSource(input: {\n source_type: \"experiment\" | \"key_result\" | \"objective\";\n source_id: string;\n}): Promise<Task[]> {\n try {\n const config = await getLLMConfig();\n if (!config.api_key) return [];\n\n // Pull source context so the LLM has something concrete to plan against.\n let contextBlock = \"\";\n let link: Partial<Task> = {};\n if (input.source_type === \"experiment\") {\n const r = await sqlite.execute(\n `SELECT title, hypothesis, funnel_stage, aarrr_lever, success_metric, baseline_value\n FROM experiments WHERE id = ?`,\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Experiment: ${row[0]}\\n` +\n `Hypothesis: ${row[1] ?? \"(none)\"}\\n` +\n `Funnel stage: ${row[2] ?? \"(unspecified)\"}\\n` +\n `AARRR lever: ${row[3] ?? \"(unspecified)\"}\\n` +\n `Success metric: ${row[4] ?? \"(unspecified)\"}\\n` +\n `Baseline: ${row[5] ?? \"(unset)\"}`;\n link = { experiment_id: input.source_id };\n } else if (input.source_type === \"key_result\") {\n const r = await sqlite.execute(\n `SELECT title, metric_source, metric_key, start_value, current_value, target_value,\n unit, objective_id\n FROM key_results WHERE id = ?`,\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Key Result: ${row[0]}\\n` +\n `Metric: ${row[1] ?? \"?\"}:${row[2] ?? \"?\"}\\n` +\n `Progress: ${row[4]} / ${row[5]} ${row[6] ?? \"\"} (started at ${row[3]})\\n` +\n `Gap to close: ${Number(row[5] ?? 0) - Number(row[4] ?? 0)} ${row[6] ?? \"\"}`;\n link = { key_result_id: input.source_id, objective_id: row[7] ?? null };\n } else {\n const r = await sqlite.execute(\n \"SELECT title, description, quarter FROM objectives WHERE id = ?\",\n [input.source_id],\n );\n if (!r.rows?.length) return [];\n const row = r.rows[0] as any[];\n contextBlock =\n `Objective: ${row[0]}\\n` +\n `Description: ${row[1] ?? \"(none)\"}\\n` +\n `Quarter: ${row[2] ?? \"(unspecified)\"}`;\n link = { objective_id: input.source_id };\n }\n\n const system =\n \"You are an operations planner breaking work into concrete tasks. \" +\n \"Return ONLY a valid JSON array (up to 5 items) of \" +\n '{\"title\":string,\"description\":string,\"priority\":\"low|medium|high|urgent\",' +\n '\"impact\":1-10,\"confidence\":1-10,\"ease\":1-10}. ' +\n \"No markdown fences, no prose outside the array.\";\n\n const user =\n `Plan up to 5 concrete next-step tasks for the following:\\n\\n${contextBlock}\\n\\n` +\n `Each task should be one focused action a single person can do in <1 week.`;\n\n let raw = \"\";\n try {\n const res = await callLLM(\n config,\n [{ role: \"user\", content: user } as any],\n { systemPrompt: system, maxTokens: 800, temperature: 0.4 },\n );\n raw = res.content || \"\";\n } catch {\n return [];\n }\n\n const generated = parseTasksJson(raw).slice(0, 5);\n if (!generated.length) return [];\n\n const inserted: Task[] = [];\n for (const g of generated) {\n if (!g.title) continue;\n const created = await createTask({\n title: String(g.title).slice(0, 200),\n description: g.description ? String(g.description).slice(0, 1000) : null,\n priority: [\"low\", \"medium\", \"high\", \"urgent\"].includes(String(g.priority))\n ? String(g.priority)\n : \"medium\",\n status: \"todo\",\n impact: g.impact != null ? clamp10(g.impact) ?? undefined : undefined,\n confidence: g.confidence != null ? clamp10(g.confidence) ?? undefined : undefined,\n ease: g.ease != null ? clamp10(g.ease) ?? undefined : undefined,\n ...link,\n } as Partial<Task>);\n inserted.push(created);\n }\n return inserted;\n } catch {\n return [];\n }\n}\n",[],[365,706],"c9f08122-5d65-11f1-8d2a-1607ee4eb77e","stats.ts","c9f08208-5d65-11f1-8d2a-1607ee4eb77e",602,["D",1780282718615],"// Part 3 data-spine — real statistics over metric_history (replaces the fabricated\n// constant-confidence correlation rules in intelligence.getCorrelations and the\n// \"typed a bigger number = win\" experiment outcomes in experiments.completeExperiment).\n//\n// Pure, dependency-free numeric helpers. Everything here is honest: if there is not\n// enough data, callers get null/insufficient — never a made-up coefficient.\n\nexport interface SeriesPoint { date: string; value: number }\n\nfunction mean(xs: number[]): number {\n return xs.length ? xs.reduce((a, b) => a + b, 0) / xs.length : 0;\n}\n\nfunction std(xs: number[], m: number): number {\n if (xs.length < 2) return 0;\n const v = xs.reduce((a, b) => a + (b - m) * (b - m), 0) / (xs.length - 1);\n return Math.sqrt(v);\n}\n\n// Pearson product-moment correlation. Returns r in [-1, 1], or 0 if undefined\n// (constant series / fewer than 2 points).\nexport function pearson(xs: number[], ys: number[]): number {\n const n = Math.min(xs.length, ys.length);\n if (n < 2) return 0;\n const mx = mean(xs.slice(0, n));\n const my = mean(ys.slice(0, n));\n let num = 0, dx = 0, dy = 0;\n for (let i = 0; i < n; i++) {\n const a = xs[i] - mx, b = ys[i] - my;\n num += a * b; dx += a * a; dy += b * b;\n }\n if (dx === 0 || dy === 0) return 0;\n const r = num / Math.sqrt(dx * dy);\n return Math.max(-1, Math.min(1, r));\n}\n\n// Spearman rank correlation — robust to non-linear monotonic relationships.\nexport function spearman(xs: number[], ys: number[]): number {\n const n = Math.min(xs.length, ys.length);\n if (n < 2) return 0;\n const rank = (arr: number[]): number[] => {\n const idx = arr.map((v, i) => [v, i]).sort((a, b) => a[0] - b[0]);\n const r = new Array(arr.length).fill(0);\n for (let i = 0; i < idx.length;) {\n let j = i;\n while (j + 1 < idx.length && idx[j + 1][0] === idx[i][0]) j++;\n const avg = (i + j) / 2 + 1; // average rank for ties (1-based)\n for (let k = i; k <= j; k++) r[idx[k][1]] = avg;\n i = j + 1;\n }\n return r;\n };\n return pearson(rank(xs.slice(0, n)), rank(ys.slice(0, n)));\n}\n\n// Two-tailed p-value for a Pearson r over n samples, via the t-distribution\n// t = r * sqrt((n-2)/(1-r^2)) with df = n-2. Uses an incomplete-beta approximation\n// of the Student-t survival function (good enough for a significance gate, not a paper).\nexport function pValueFromPearson(r: number, n: number): number {\n if (n < 3) return 1;\n const rr = Math.min(0.999999, Math.abs(r));\n const df = n - 2;\n const t = rr * Math.sqrt(df / (1 - rr * rr));\n return studentTTwoTailed(t, df);\n}\n\n// Student-t two-tailed p-value via the regularized incomplete beta function.\nfunction studentTTwoTailed(t: number, df: number): number {\n const x = df / (df + t * t);\n // P(|T| > t) = I_x(df/2, 1/2)\n const p = betai(df / 2, 0.5, x);\n return Math.max(0, Math.min(1, p));\n}\n\n// Regularized incomplete beta I_x(a,b) via continued fraction (Numerical Recipes).\nfunction betai(a: number, b: number, x: number): number {\n if (x <= 0) return 0;\n if (x >= 1) return 1;\n const lbeta = gammaln(a + b) - gammaln(a) - gammaln(b);\n const front = Math.exp(lbeta + a * Math.log(x) + b * Math.log(1 - x));\n if (x < (a + 1) / (a + b + 2)) return front * betacf(a, b, x) / a;\n return 1 - front * betacf(b, a, 1 - x) / b;\n}\n\nfunction betacf(a: number, b: number, x: number): number {\n const MAXIT = 200, EPS = 3e-12, FPMIN = 1e-300;\n let qab = a + b, qap = a + 1, qam = a - 1;\n let c = 1, d = 1 - qab * x / qap;\n if (Math.abs(d) < FPMIN) d = FPMIN;\n d = 1 / d; let h = d;\n for (let m = 1; m <= MAXIT; m++) {\n const m2 = 2 * m;\n let aa = m * (b - m) * x / ((qam + m2) * (a + m2));\n d = 1 + aa * d; if (Math.abs(d) < FPMIN) d = FPMIN;\n c = 1 + aa / c; if (Math.abs(c) < FPMIN) c = FPMIN;\n d = 1 / d; h *= d * c;\n aa = -(a + m) * (qab + m) * x / ((a + m2) * (qap + m2));\n d = 1 + aa * d; if (Math.abs(d) < FPMIN) d = FPMIN;\n c = 1 + aa / c; if (Math.abs(c) < FPMIN) c = FPMIN;\n d = 1 / d; const del = d * c; h *= del;\n if (Math.abs(del - 1) < EPS) break;\n }\n return h;\n}\n\nfunction gammaln(xx: number): number {\n const cof = [76.18009172947146, -86.50532032941677, 24.01409824083091,\n -1.231739572450155, 0.1208650973866179e-2, -0.5395239384953e-5];\n let x = xx, y = xx;\n let tmp = x + 5.5;\n tmp -= (x + 0.5) * Math.log(tmp);\n let ser = 1.000000000190015;\n for (let j = 0; j < 6; j++) { y += 1; ser += cof[j] / y; }\n return -tmp + Math.log(2.5066282746310005 * ser / x);\n}\n\n// Align two daily series by date (inner join on the date string) → parallel value arrays.\nexport function alignSeries(a: SeriesPoint[], b: SeriesPoint[]): { xs: number[]; ys: number[] } {\n const mb = new Map(b.map((p) => [p.date, p.value]));\n const xs: number[] = [], ys: number[] = [];\n for (const p of a) {\n if (mb.has(p.date)) { xs.push(p.value); ys.push(mb.get(p.date) as number); }\n }\n return { xs, ys };\n}\n\n// Best lagged correlation: tries lags 0..maxLag where series B trails series A by `lag`\n// days (i.e. a change in A precedes a change in B). Returns the lag with the largest |r|,\n// the coefficient, its p-value, and the paired-sample count n at that lag.\nexport function bestLaggedCorrelation(\n a: SeriesPoint[],\n b: SeriesPoint[],\n maxLag = 7,\n method: \"pearson\" | \"spearman\" = \"pearson\",\n): { r: number; lag: number; n: number; p: number } {\n let best = { r: 0, lag: 0, n: 0, p: 1 };\n const shiftDate = (d: string, days: number): string => {\n const t = new Date(d + \"T00:00:00Z\").getTime() + days * 86400000;\n return new Date(t).toISOString().slice(0, 10);\n };\n for (let lag = 0; lag <= maxLag; lag++) {\n // Shift B back by `lag` so A[date] aligns with B[date+lag].\n const bShift = b.map((p) => ({ date: shiftDate(p.date, -lag), value: p.value }));\n const { xs, ys } = alignSeries(a, bShift);\n if (xs.length < 3) continue;\n const r = method === \"spearman\" ? spearman(xs, ys) : pearson(xs, ys);\n if (Math.abs(r) > Math.abs(best.r)) {\n best = { r, lag, n: xs.length, p: pValueFromPearson(r, xs.length) };\n }\n }\n return best;\n}\n\n// Two-proportion z-test for experiment significance (control vs variant conversion rates).\n// Returns z, the two-tailed p-value, and the pooled rates. Honest: needs both arms' counts.\nexport function twoProportionZTest(\n c1: number, n1: number, c2: number, n2: number,\n): { z: number; p: number; rate1: number; rate2: number; n: number } | null {\n if (n1 <= 0 || n2 <= 0) return null;\n const p1 = c1 / n1, p2 = c2 / n2;\n const pPool = (c1 + c2) / (n1 + n2);\n const se = Math.sqrt(pPool * (1 - pPool) * (1 / n1 + 1 / n2));\n if (se === 0) return null;\n const z = (p2 - p1) / se;\n const p = 2 * (1 - normalCdf(Math.abs(z)));\n return { z, p: Math.max(0, Math.min(1, p)), rate1: p1, rate2: p2, n: n1 + n2 };\n}\n\n// Standard normal CDF (Abramowitz & Stegun 7.1.26 erf approximation).\nexport function normalCdf(x: number): number {\n const t = 1 / (1 + 0.2316419 * Math.abs(x));\n const d = 0.3989422804014327 * Math.exp(-x * x / 2);\n const prob = d * t * (0.319381530 + t * (-0.356563782 + t * (1.781477937 + t * (-1.821255978 + t * 1.330274429))));\n return x > 0 ? 1 - prob : prob;\n}\n\n// Minimum paired samples before a correlation is allowed to surface, and the |r| floor.\nexport const MIN_CORR_SAMPLES = 8;\nexport const MIN_CORR_ABS_R = 0.5;\nexport const MAX_CORR_P = 0.1;\n\n// Classify |r| into the app's strength buckets.\nexport function corrStrength(absR: number): \"weak\" | \"moderate\" | \"strong\" {\n if (absR >= 0.7) return \"strong\";\n if (absR >= 0.5) return \"moderate\";\n return \"weak\";\n}\n",[],[365,714],"ea50738e-5d6d-11f1-95b0-1607ee4eb77e","skills.ts","ea507456-5d6d-11f1-95b0-1607ee4eb77e",643,["D",1780286208905],"// Part 1 — Skills service. Mirrors knowledge.ts style. Bodies come from the bundled\n// catalog (skills/index.ts, read from memory); the `skills` DB table holds only per-skill\n// STATE (enabled/pinned) + user-created custom skills. getEffectiveSkills() merges them.\nimport { sqlite } from \"../db.ts\";\nimport {\n BUILTIN_SKILLS, getSkillDef, getSkillBody, listReferences, getReference,\n type SkillDef,\n} from \"../skills/index.ts\";\nimport { getCompanyContextString } from \"./company.ts\";\n\nconst APPLY_CAP = 12000; // max chars of a skill body returned by applySkillContext\nconst INDEX_CAP = 6500; // max chars of the prompt-injected skills index\n\nexport interface EffectiveSkill {\n name: string;\n category: string;\n description: string;\n version: string;\n enabled: boolean;\n pinned: boolean;\n source: \"builtin\" | \"user\";\n has_body: boolean;\n}\n\ninterface SkillStateRow { enabled: number; pinned: number; source: string; category: string | null; description: string | null; body: string | null }\n\nasync function loadState(): Promise<Map<string, SkillStateRow>> {\n const map = new Map<string, SkillStateRow>();\n try {\n const r = await sqlite.execute(\"SELECT name, enabled, pinned, source, category, description, body FROM skills\");\n for (const row of (r.rows || []) as any[]) {\n const name = String(row[0]);\n map.set(name, {\n enabled: Number(row[1]), pinned: Number(row[2]), source: String(row[3] || \"builtin\"),\n category: row[4] != null ? String(row[4]) : null,\n description: row[5] != null ? String(row[5]) : null,\n body: row[6] != null ? String(row[6]) : null,\n });\n }\n } catch { /* table missing → all builtins default-enabled */ }\n return map;\n}\n\n// Merge builtin catalog + DB state + custom (source='user') skills.\nexport async function getEffectiveSkills(): Promise<EffectiveSkill[]> {\n const state = await loadState();\n const out: EffectiveSkill[] = [];\n const seen = new Set<string>();\n for (const b of BUILTIN_SKILLS) {\n const s = state.get(b.name);\n out.push({\n name: b.name, category: b.category, description: b.description, version: b.version,\n enabled: s ? s.enabled !== 0 : true,\n pinned: s ? s.pinned === 1 : false,\n source: \"builtin\", has_body: true,\n });\n seen.add(b.name);\n }\n for (const [name, s] of state) {\n if (seen.has(name) || s.source !== \"user\") continue;\n out.push({\n name, category: s.category || \"Custom\", description: s.description || \"\", version: \"1.0.0\",\n enabled: s.enabled !== 0, pinned: s.pinned === 1, source: \"user\", has_body: !!s.body,\n });\n }\n return out;\n}\n\n// Grouped-by-category list (no body) for the Playbooks page + GET /api/skills.\nexport async function listSkills(): Promise<{\n categories: Array<{ category: string; skills: EffectiveSkill[] }>;\n total: number; enabled: number;\n}> {\n const all = await getEffectiveSkills();\n const byCat = new Map<string, EffectiveSkill[]>();\n for (const s of all) {\n if (!byCat.has(s.category)) byCat.set(s.category, []);\n byCat.get(s.category)!.push(s);\n }\n const categories = [...byCat.entries()]\n .map(([category, skills]) => ({ category, skills: skills.sort((a, b) => a.name.localeCompare(b.name)) }))\n .sort((a, b) => a.category.localeCompare(b.category));\n return { categories, total: all.length, enabled: all.filter((s) => s.enabled).length };\n}\n\n// Full skill (body + reference list). Custom bodies come from the DB; builtin from catalog.\nexport async function getSkill(name: string): Promise<\n (EffectiveSkill & { body: string; references: string[] }) | null\n> {\n const eff = (await getEffectiveSkills()).find((s) => s.name === name);\n if (!eff) return null;\n let body = getSkillBody(name) ?? \"\";\n let refs = listReferences(name);\n if (eff.source === \"user\") {\n const state = await loadState();\n body = state.get(name)?.body || \"\";\n refs = [];\n }\n return { ...eff, body, references: refs };\n}\n\n// Compact index of ENABLED skills for prompt injection: \"- name — first sentence [Category]\".\nexport async function getSkillsIndex(opts: { maxChars?: number } = {}): Promise<string> {\n const cap = opts.maxChars ?? INDEX_CAP;\n const all = (await getEffectiveSkills()).filter((s) => s.enabled);\n if (!all.length) return \"\";\n all.sort((a, b) => (b.pinned ? 1 : 0) - (a.pinned ? 1 : 0) || a.category.localeCompare(b.category) || a.name.localeCompare(b.name));\n const lines: string[] = [];\n let used = 0;\n for (const s of all) {\n const first = (s.description.split(/(?<=[.!?])\\s/)[0] || s.description).slice(0, 160).trim();\n const line = `- ${s.name} — ${first} [${s.category}]`;\n if (used + line.length + 1 > cap) break;\n lines.push(line); used += line.length + 1;\n }\n return \"MARKETING SKILLS INDEX (call apply_skill with the skill name when a request matches one):\\n\" + lines.join(\"\\n\");\n}\n\n// The string the apply_skill tool returns: brand/company context + the full playbook body.\nexport async function applySkillContext(name: string): Promise<string | null> {\n const def = getSkillDef(name);\n let body = def?.body ?? null;\n if (!body) {\n const state = await loadState();\n body = state.get(name)?.body ?? null;\n }\n if (!body) return null;\n let brand = \"\";\n try { brand = await getCompanyContextString(1500); } catch { /* optional */ }\n const head = brand ? brand + \"\\n\\n---\\n\\n\" : \"\";\n return (head + body).slice(0, APPLY_CAP + head.length);\n}\n\n// Reference doc fetch for the get_skill_reference tool.\nexport function getSkillReference(name: string, ref: string): string | null {\n return getReference(name, ref);\n}\nexport function getSkillReferenceList(name: string): string[] {\n return listReferences(name);\n}\n\nexport async function setSkillState(name: string, patch: { enabled?: boolean; pinned?: boolean }): Promise<boolean> {\n const isBuiltin = !!getSkillDef(name);\n const state = await loadState();\n const existing = state.get(name);\n if (!isBuiltin && !existing) return false; // unknown skill\n const cat = existing?.category ?? getSkillDef(name)?.category ?? null;\n const desc = existing?.description ?? getSkillDef(name)?.description ?? null;\n const enabled = patch.enabled != null ? (patch.enabled ? 1 : 0) : (existing ? existing.enabled : 1);\n const pinned = patch.pinned != null ? (patch.pinned ? 1 : 0) : (existing ? existing.pinned : 0);\n await sqlite.execute(\n `INSERT INTO skills (name, enabled, pinned, source, category, description, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, datetime('now'))\n ON CONFLICT(name) DO UPDATE SET enabled=excluded.enabled, pinned=excluded.pinned, updated_at=datetime('now')`,\n [name, enabled, pinned, existing?.source ?? (isBuiltin ? \"builtin\" : \"user\"), cat, desc],\n );\n return true;\n}\n\nexport async function createCustomSkill(data: { name: string; category?: string; description?: string; body: string }): Promise<{ ok: boolean; error?: string }> {\n const name = String(data.name || \"\").trim().toLowerCase().replace(/\\s+/g, \"-\");\n if (!name) return { ok: false, error: \"name required\" };\n if (getSkillDef(name)) return { ok: false, error: \"name collides with a builtin skill\" };\n if (!data.body || data.body.length < 10) return { ok: false, error: \"body too short\" };\n if (data.body.length > 60000) return { ok: false, error: \"body too large (cap 60KB)\" };\n await sqlite.execute(\n `INSERT INTO skills (name, enabled, pinned, source, category, description, body, updated_at)\n VALUES (?, 1, 0, 'user', ?, ?, ?, datetime('now'))\n ON CONFLICT(name) DO UPDATE SET category=excluded.category, description=excluded.description, body=excluded.body, updated_at=datetime('now')`,\n [name, data.category || \"Custom\", (data.description || \"\").slice(0, 1000), data.body],\n );\n return { ok: true };\n}\n",[],[365,722],"2af1af84-5d69-11f1-b581-1607ee4eb77e","signals.ts","9808cfb4-589e-11f1-9c4f-ee650bb23af1",623,["D",1780284169846],"// Intent signals — captures and scores buying-intent events per lead so the team can\n// triage hot leads in real time. Signals are typed (email_reply, pricing_keyword, etc.)\n// and each carries both a strength (1-3) and a score (5-50). Hot leads = leads whose\n// 30-day unactioned signal score exceeds 50.\n//\n// Sourced opportunistically from Gmail (via intelligence.getLeadScores) so the table\n// stays fresh without requiring a separate webhook ingestion pipeline.\nimport { sqlite } from \"../db.ts\";\n\n// ─── Weights ─────────────────────────────────────────────────────────────────\n// strength: 1 (weak) / 2 (medium) / 3 (strong) — used by UI badge.\n// score: numeric contribution to hot-lead score (summed over 30d window).\nexport const SIGNAL_WEIGHTS: Record<string, { strength: number; score: number }> = {\n email_open: { strength: 1, score: 5 },\n email_reply: { strength: 3, score: 25 },\n pricing_keyword: { strength: 3, score: 30 },\n demo_request: { strength: 3, score: 40 },\n linkedin_view: { strength: 1, score: 10 },\n calendar_event: { strength: 3, score: 50 },\n content_download: { strength: 2, score: 20 },\n};\n\nexport interface IntentSignal {\n id: string;\n lead_email: string;\n lead_name: string | null;\n company: string | null;\n signal_type: string;\n signal_source: string;\n signal_strength: number;\n raw_data: any;\n detected_at: string;\n actioned: number;\n actioned_at: string | null;\n}\n\nfunction rowToSignal(columns: string[], row: any[]): IntentSignal {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n let raw: any = null;\n try { raw = obj.raw_data ? JSON.parse(String(obj.raw_data)) : null; } catch { raw = String(obj.raw_data ?? \"\"); }\n return {\n id: String(obj.id),\n lead_email: String(obj.lead_email),\n lead_name: obj.lead_name ?? null,\n company: obj.company ?? null,\n signal_type: String(obj.signal_type),\n signal_source: String(obj.signal_source ?? \"\"),\n signal_strength: Number(obj.signal_strength) || 1,\n raw_data: raw,\n detected_at: String(obj.detected_at),\n actioned: Number(obj.actioned) || 0,\n actioned_at: obj.actioned_at ?? null,\n };\n}\n\n// ─── CRUD ─────────────────────────────────────────────────────────────────────\n\nexport async function listSignals(filters?: { lead_email?: string; actioned?: boolean }): Promise<IntentSignal[]> {\n // Best-effort refresh first — never fails the read if integrations are down.\n try { await syncSignalsFromIntegrations(); } catch { /* non-critical */ }\n\n const where: string[] = [];\n const args: any[] = [];\n if (filters?.lead_email) { where.push(\"lead_email = ?\"); args.push(filters.lead_email); }\n if (typeof filters?.actioned === \"boolean\") { where.push(\"actioned = ?\"); args.push(filters.actioned ? 1 : 0); }\n const sql = `SELECT id, lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned, actioned_at\n FROM intent_signals\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY detected_at DESC LIMIT 100`;\n try {\n const r = await sqlite.execute(sql, args);\n return (r.rows || []).map((row) => rowToSignal(r.columns, row as any[]));\n } catch { return []; }\n}\n\nexport async function recordSignal(data: {\n lead_email: string;\n lead_name?: string;\n company?: string;\n signal_type: string;\n signal_source?: string;\n source?: string; // alias accepted for convenience\n raw_data?: any;\n score?: number; // ignored — strength is derived from SIGNAL_WEIGHTS\n workspace?: string; // ignored at signal level\n}): Promise<IntentSignal> {\n const weight = SIGNAL_WEIGHTS[data.signal_type] ?? { strength: 1, score: 5 };\n const rawJson = data.raw_data != null ? JSON.stringify(data.raw_data) : null;\n const now = new Date().toISOString();\n const signal_source = data.signal_source ?? data.source ?? \"manual\";\n await sqlite.execute(\n `INSERT INTO intent_signals (lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, 0)`,\n [data.lead_email, data.lead_name ?? null, data.company ?? null,\n data.signal_type, signal_source, weight.strength, rawJson, now]\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return {\n id, lead_email: data.lead_email, lead_name: data.lead_name ?? null, company: data.company ?? null,\n signal_type: data.signal_type, signal_source,\n signal_strength: weight.strength, raw_data: data.raw_data ?? null,\n detected_at: now, actioned: 0, actioned_at: null,\n };\n}\n\nexport async function actionSignal(id: string): Promise<IntentSignal | null> {\n const now = new Date().toISOString();\n await sqlite.execute(\n \"UPDATE intent_signals SET actioned = 1, actioned_at = ? WHERE id = ?\",\n [now, id]\n );\n const r = await sqlite.execute(\n `SELECT id, lead_email, lead_name, company, signal_type, signal_source,\n signal_strength, raw_data, detected_at, actioned, actioned_at\n FROM intent_signals WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) return null;\n return rowToSignal(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Hot leads aggregation ───────────────────────────────────────────────────\n// Group unactioned signals from the last 30 days by lead_email, sum weighted\n// scores per signal_type, return leads with total > 50 sorted desc.\n\nexport interface HotLead {\n email: string;\n name: string;\n company: string;\n score: number;\n top_signals: string[];\n signal_count: number;\n last_detected_at: string;\n}\n\nexport async function getHotLeads(): Promise<HotLead[]> {\n let rows: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT lead_email, lead_name, company, signal_type, detected_at\n FROM intent_signals\n WHERE actioned = 0\n AND detected_at >= datetime('now','-30 days')\n ORDER BY detected_at DESC`\n );\n rows = (r.rows || []) as any[];\n } catch { rows = []; }\n\n // Aggregate by email.\n type Agg = {\n email: string;\n name: string;\n company: string;\n score: number;\n type_counts: Map<string, number>;\n signal_count: number;\n last_detected_at: string;\n };\n const byEmail = new Map<string, Agg>();\n for (const row of rows) {\n const email = String(row[0]);\n const name = String(row[1] ?? email.split(\"@\")[0]);\n const company = String(row[2] ?? \"\");\n const type = String(row[3]);\n const detected = String(row[4]);\n const weight = SIGNAL_WEIGHTS[type]?.score ?? 5;\n let agg = byEmail.get(email);\n if (!agg) {\n agg = { email, name, company, score: 0, type_counts: new Map(), signal_count: 0, last_detected_at: detected };\n byEmail.set(email, agg);\n }\n agg.score += weight;\n agg.signal_count += 1;\n agg.type_counts.set(type, (agg.type_counts.get(type) || 0) + 1);\n if (detected > agg.last_detected_at) agg.last_detected_at = detected;\n }\n\n const hot: HotLead[] = [];\n for (const agg of byEmail.values()) {\n if (agg.score <= 50) continue;\n // top_signals: 3 most frequent signal types for this lead\n const top = Array.from(agg.type_counts.entries())\n .sort((a, b) => b[1] - a[1])\n .slice(0, 3)\n .map(([t]) => t);\n hot.push({\n email: agg.email,\n name: agg.name,\n company: agg.company,\n score: agg.score,\n top_signals: top,\n signal_count: agg.signal_count,\n last_detected_at: agg.last_detected_at,\n });\n }\n hot.sort((a, b) => b.score - a.score);\n return hot;\n}\n\nexport async function getHotLeadsCount(): Promise<number> {\n const leads = await getHotLeads();\n return leads.length;\n}\n\n// ─── Sync from intelligence layer ────────────────────────────────────────────\n// Pulls fresh lead scores from intelligence.getLeadScores and converts MQL/SQL\n// behavior into intent_signals rows. Idempotent-ish: skips signals that already\n// exist (same email + type within the last 24h) to avoid runaway duplicates.\n\nexport async function syncSignalsFromIntegrations(): Promise<{ recorded: number; skipped: number }> {\n let recorded = 0;\n let skipped = 0;\n try {\n const { getLeadScores } = await import(\"./intelligence.ts\");\n const leads = await getLeadScores();\n for (const lead of leads) {\n if ((lead.mql_score ?? 0) < 25) continue;\n // Determine which signal types to emit based on what the lead's breakdown shows.\n const wants: Array<{ type: string; source: string }> = [];\n const sigs = (lead.sql_signals || []) as string[];\n const bk = (lead.signal_breakdown || {}) as Record<string, number>;\n\n if (sigs.includes(\"reply_to_outreach_email\") || sigs.includes(\"replied_to_thread\")) {\n wants.push({ type: \"email_reply\", source: lead.source || \"gmail\" });\n }\n if (bk.visits_pricing || bk.gmail_pricing) {\n wants.push({ type: \"pricing_keyword\", source: lead.source || \"gmail\" });\n }\n if (sigs.includes(\"requested_demo_or_pricing\") || sigs.includes(\"requested_demo_or_trial\") || sigs.includes(\"requested_proposal\")) {\n wants.push({ type: \"demo_request\", source: lead.source || \"gmail\" });\n }\n if (bk.downloads_content) {\n wants.push({ type: \"content_download\", source: lead.source || \"gmail\" });\n }\n if (sigs.includes(\"booked_discovery_call\")) {\n wants.push({ type: \"calendar_event\", source: \"outlook\" });\n }\n if (wants.length === 0 && (lead.mql_score ?? 0) >= 40) {\n // High MQL score but no specific signal — log a generic email_open.\n wants.push({ type: \"email_open\", source: lead.source || \"gmail\" });\n }\n\n for (const w of wants) {\n // Dedup: skip if this email already has this signal_type within the last 24h.\n try {\n const dup = await sqlite.execute(\n `SELECT id FROM intent_signals\n WHERE lead_email = ? AND signal_type = ?\n AND detected_at >= datetime('now','-1 day') LIMIT 1`,\n [lead.email, w.type]\n );\n if (dup.rows?.length) { skipped++; continue; }\n } catch { /* tolerate query failure, attempt insert */ }\n\n try {\n await recordSignal({\n lead_email: lead.email,\n lead_name: lead.name,\n company: lead.company,\n signal_type: w.type,\n signal_source: w.source,\n raw_data: { mql_score: lead.mql_score, stage: lead.stage },\n });\n recorded++;\n } catch { /* swallow individual insert error */ }\n }\n }\n } catch {\n // intelligence module unavailable or threw — safe no-op.\n }\n return { recorded, skipped };\n}\n",[],[365,730],"2a9d71a8-5d69-11f1-a87a-1607ee4eb77e","propagate.ts","514dfcee-5a69-11f1-bcd7-ee650bb23af1",622,["D",1780284169295],"// Closed-loop propagation — when work is EXECUTED (a task is marked done or an\n// experiment is completed), reflect it in the MEASURE layer so execution and\n// measurement stop being decoupled:\n// 1) bump an internal \"executed work\" counter (metric_history, integration='internal')\n// 2) drop a marker onto the unified calendar (calendar_entries → 'custom' source)\n// 3) refresh the linked objective's progress (mean of its KR progress)\n//\n// Self-contained: imports ONLY sqlite to avoid import cycles with okrs/experiments/tasks.\nimport { sqlite } from \"../db.ts\";\n\nconst today = () => new Date().toISOString().slice(0, 10);\n\nasync function bumpExecutedWork(metricKey: string): Promise<void> {\n try {\n await sqlite.execute(\n \"INSERT INTO metric_history (integration, metric_key, value) VALUES ('internal', ?, 1)\",\n [metricKey],\n );\n } catch { /* non-critical */ }\n}\n\nasync function addCalendarMarker(title: string, workspace: string | null, color: string): Promise<void> {\n try {\n await sqlite.execute(\n `INSERT INTO calendar_entries (title, event_date, event_type, workspace, color, all_day)\n VALUES (?, ?, 'completion', ?, ?, 1)`,\n [title, today(), workspace ?? \"marketing\", color],\n );\n } catch { /* non-critical */ }\n}\n\n// Recompute one objective's progress = mean of its KR progress (matches okrs.computeObjectiveProgress).\nasync function refreshObjectiveProgress(objectiveId: string | number | null): Promise<void> {\n if (objectiveId == null) return;\n try {\n const r = await sqlite.execute(\n \"SELECT progress FROM key_results WHERE objective_id = ?\",\n [objectiveId],\n );\n const rows = (r.rows || []) as any[];\n if (!rows.length) return;\n const mean = Math.round(rows.reduce((a, row) => a + (Number(row[0]) || 0), 0) / rows.length);\n await sqlite.execute(\n \"UPDATE objectives SET progress = ?, updated_at = datetime('now') WHERE id = ?\",\n [mean, objectiveId],\n );\n } catch { /* non-critical */ }\n}\n\n// A task links to an objective directly or via its key_result.\nasync function objectiveForTask(t: { objective_id: any; key_result_id: any }): Promise<string | null> {\n if (t.objective_id != null) return String(t.objective_id);\n if (t.key_result_id != null) {\n try {\n const r = await sqlite.execute(\"SELECT objective_id FROM key_results WHERE id = ?\", [t.key_result_id]);\n const row = ((r.rows || [])[0]) as any[] | undefined;\n if (row && row[0] != null) return String(row[0]);\n } catch { /* */ }\n }\n return null;\n}\n\nexport async function propagateTaskDone(task: {\n title: string; workspace: string | null; objective_id: any; key_result_id: any;\n}): Promise<void> {\n await bumpExecutedWork(\"tasks_completed\");\n await addCalendarMarker(\"Done: \" + (task.title || \"task\"), task.workspace, \"green\");\n await refreshObjectiveProgress(await objectiveForTask(task));\n}\n\nexport async function propagateExperimentComplete(exp: {\n title: string; outcome: string | null; objective_id: any;\n}): Promise<void> {\n await bumpExecutedWork(\"experiments_completed\");\n if (exp.outcome === \"win\") await bumpExecutedWork(\"experiments_won\");\n await addCalendarMarker(\n \"Experiment \" + (exp.outcome || \"done\") + \": \" + (exp.title || \"\"),\n null,\n exp.outcome === \"win\" ? \"secondary\" : \"tertiary\",\n );\n await refreshObjectiveProgress(exp.objective_id != null ? String(exp.objective_id) : null);\n}\n\n// Executed-work counts over a window — read by intelligence for execution-aware KPIs.\nexport async function getExecutedWorkCounts(\n days = 30,\n): Promise<{ tasks_completed: number; experiments_completed: number; experiments_won: number }> {\n const out = { tasks_completed: 0, experiments_completed: 0, experiments_won: 0 };\n try {\n const r = await sqlite.execute(\n `SELECT metric_key, COUNT(*) FROM metric_history\n WHERE integration = 'internal' AND recorded_at >= datetime('now', ?)\n GROUP BY metric_key`,\n [`-${days} days`],\n );\n for (const row of (r.rows || []) as any[]) {\n const k = String(row[0]);\n if (k in out) (out as any)[k] = Number(row[1]) || 0;\n }\n } catch { /* */ }\n return out;\n}\n",[],[365,738],"2a2387bc-5d69-11f1-b2ca-1607ee4eb77e","okrs.ts","96e84dee-589e-11f1-8829-ee650bb23af1",621,["D",1780284168488],"// OKRs — Objectives + Key Results. Phase 1 Framework Layer.\n// Objectives roll up KR progress; KR current_value auto-syncs from the funnel\n// summary when metric_source is a connected integration.\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\nimport { getAttributionSummary } from \"./aio.ts\";\n\n// Current quarter as \"YYYY-Qn\" (runs on val.town where Date is available).\nfunction currentQuarter(): string {\n const d = new Date();\n return d.getFullYear() + \"-Q\" + (Math.floor(d.getMonth() / 3) + 1);\n}\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport interface KeyResult {\n id: string;\n objective_id: string;\n title: string;\n metric_source: string | null;\n metric_key: string | null;\n start_value: number;\n current_value: number;\n target_value: number;\n unit: string | null;\n confidence: number; // 0..1\n progress: number; // 0..100\n updated_at: string;\n}\n\nexport interface Objective {\n id: string;\n title: string;\n description: string | null;\n quarter: string | null;\n workspace: string | null;\n owner: string | null;\n status: string; // draft | active | done | archived\n progress: number; // 0..100\n created_at: string;\n updated_at: string;\n key_results: KeyResult[];\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp = (n: number, lo: number, hi: number) =>\n Math.max(lo, Math.min(hi, n));\n\nfunction computeKRProgress(kr: KeyResult): number {\n const start = Number(kr.start_value) || 0;\n const cur = Number(kr.current_value) || 0;\n const target = Number(kr.target_value) || 0;\n if (target === start) {\n // Degenerate: target == start. If we're at/above target, call it 100%.\n return cur >= target ? 100 : 0;\n }\n const pct = ((cur - start) / (target - start)) * 100;\n return Math.round(clamp(pct, 0, 100));\n}\n\nfunction computeObjectiveProgress(krs: KeyResult[]): number {\n if (!krs.length) return 0;\n const sum = krs.reduce((acc, k) => acc + (Number(k.progress) || 0), 0);\n return Math.round(sum / krs.length);\n}\n\n// Pulls live integration value for a KR using the unified funnel summary.\n// Mutates kr.current_value + kr.progress in place. Falls through silently on\n// any error so we never break list endpoints because the funnel fetch hiccuped.\nfunction syncKRFromFunnel(\n kr: KeyResult,\n funnelMetrics: Array<{ integration: string; metric: string; value: number }>,\n): void {\n if (!kr.metric_source || !kr.metric_key) {\n kr.progress = computeKRProgress(kr);\n return;\n }\n const hit = funnelMetrics.find(\n (m) => m.integration === kr.metric_source && m.metric === kr.metric_key,\n );\n if (hit && typeof hit.value === \"number\" && Number.isFinite(hit.value)) {\n kr.current_value = Number(hit.value);\n }\n kr.progress = computeKRProgress(kr);\n}\n\nfunction rowToKR(columns: string[], row: any[]): KeyResult {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n objective_id: String(obj.objective_id),\n title: String(obj.title ?? \"\"),\n metric_source: obj.metric_source ?? null,\n metric_key: obj.metric_key ?? null,\n start_value: Number(obj.start_value) || 0,\n current_value: Number(obj.current_value) || 0,\n target_value: Number(obj.target_value) || 0,\n unit: obj.unit ?? null,\n confidence: Number(obj.confidence ?? 0.5),\n progress: Number(obj.progress) || 0,\n updated_at: String(obj.updated_at ?? \"\"),\n };\n}\n\nfunction rowToObjective(columns: string[], row: any[]): Objective {\n const obj = rowToObj(columns, row);\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n quarter: obj.quarter ?? null,\n workspace: obj.workspace ?? null,\n owner: obj.owner ?? null,\n status: String(obj.status ?? \"draft\"),\n progress: Number(obj.progress) || 0,\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n key_results: [],\n };\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listObjectives(\n filters: { workspace?: string; quarter?: string } = {},\n): Promise<Objective[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.workspace) { where.push(\"workspace = ?\"); args.push(filters.workspace); }\n if (filters.quarter) { where.push(\"quarter = ?\"); args.push(filters.quarter); }\n const sql = `SELECT id, title, description, quarter, workspace, owner, status, progress,\n created_at, updated_at\n FROM objectives\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}\n ORDER BY created_at DESC`;\n const r = await sqlite.execute(sql, args);\n const objectives = (r.rows || []).map(\n (row) => rowToObjective(r.columns, row as any[]),\n );\n if (objectives.length === 0) return [];\n\n // Pull KRs for these objectives in one query.\n const ids = objectives.map((o) => o.id);\n const placeholders = ids.map(() => \"?\").join(\",\");\n const krResult = await sqlite.execute(\n `SELECT id, objective_id, title, metric_source, metric_key, start_value,\n current_value, target_value, unit, confidence, progress, updated_at\n FROM key_results\n WHERE objective_id IN (${placeholders})`,\n ids,\n );\n const krs = (krResult.rows || []).map(\n (row) => rowToKR(krResult.columns, row as any[]),\n );\n\n // Refresh KR current_value from live funnel data for DISPLAY ONLY — no DB writes here.\n // Persisting is the job of syncObjectivesFromFunnel() (cron + POST /api/okrs/sync), so a\n // plain GET stays idempotent (was previously writing key_results + objectives on every read).\n const funnelMetrics = await loadFunnelMetrics();\n for (const kr of krs) syncKRFromFunnel(kr, funnelMetrics);\n\n // Attach KRs + recompute objective progress (in memory).\n const byObj = new Map<string, KeyResult[]>();\n for (const kr of krs) {\n if (!byObj.has(kr.objective_id)) byObj.set(kr.objective_id, []);\n byObj.get(kr.objective_id)!.push(kr);\n }\n for (const obj of objectives) {\n obj.key_results = byObj.get(obj.id) ?? [];\n obj.progress = computeObjectiveProgress(obj.key_results);\n }\n return objectives;\n } catch {\n return [];\n }\n}\n\n// Flatten live funnel data into {integration, metric, value} rows (read-only).\nasync function loadFunnelMetrics(): Promise<Array<{ integration: string; metric: string; value: number }>> {\n try {\n const summary = await getFunnelSummary();\n const rows = summary.stages.flatMap((s) =>\n s.metrics.map((m: any) => ({\n integration: m.integration,\n metric: m.metric,\n value: Number(m.value) || 0,\n })),\n );\n // Revenue wiring: expose CRM/attribution revenue as a funnel metric so a KR with\n // metric_source 'attribution' + metric_key 'revenue' (e.g. the $80K Q2 KR) can actually\n // progress. Won deals call recordConversion → utm_conversions → getAttributionSummary.total_value.\n try {\n const attr = await getAttributionSummary();\n rows.push({ integration: \"attribution\", metric: \"revenue\", value: Number(attr.total_value) || 0 });\n } catch { /* attribution optional */ }\n return rows;\n } catch {\n return [];\n }\n}\n\n// Write path: refresh KR current_value/progress from live funnel data AND persist it, then\n// recompute + persist each objective's progress. Called by the execution cron and by\n// POST /api/okrs/sync — NEVER on a plain GET. Returns how many rows were updated.\nexport async function syncObjectivesFromFunnel(\n filters: { workspace?: string; quarter?: string } = {},\n): Promise<{ objectives: number; key_results: number }> {\n try {\n const objectives = await listObjectives(filters); // read-only; attaches KRs + in-memory funnel sync\n if (objectives.length === 0) return { objectives: 0, key_results: 0 };\n const now = new Date().toISOString();\n let krCount = 0;\n for (const obj of objectives) {\n for (const kr of obj.key_results) {\n try {\n await sqlite.execute(\n \"UPDATE key_results SET current_value = ?, progress = ?, updated_at = ? WHERE id = ?\",\n [kr.current_value, kr.progress, now, kr.id],\n );\n krCount++;\n } catch { /* non-critical */ }\n }\n try {\n await sqlite.execute(\n \"UPDATE objectives SET progress = ?, updated_at = ? WHERE id = ?\",\n [obj.progress, now, obj.id],\n );\n } catch { /* non-critical */ }\n }\n return { objectives: objectives.length, key_results: krCount };\n } catch {\n return { objectives: 0, key_results: 0 };\n }\n}\n\n// ─── Write (Objectives) ──────────────────────────────────────────────────────\n\nexport async function createObjective(\n data: Partial<Objective>,\n): Promise<Objective> {\n const now = new Date().toISOString();\n await sqlite.execute(\n `INSERT INTO objectives\n (title, description, quarter, workspace, owner, status, progress, created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, 0, ?, ?)`,\n [\n data.title ?? \"Untitled Objective\",\n data.description ?? null,\n data.quarter ?? currentQuarter(),\n data.workspace ?? null,\n data.owner ?? null,\n data.status ?? \"draft\",\n now,\n now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return {\n id,\n title: data.title ?? \"Untitled Objective\",\n description: data.description ?? null,\n quarter: data.quarter ?? currentQuarter(),\n workspace: data.workspace ?? null,\n owner: data.owner ?? null,\n status: data.status ?? \"draft\",\n progress: 0,\n created_at: now,\n updated_at: now,\n key_results: [],\n };\n}\n\nexport async function updateObjective(\n id: string,\n data: Partial<Objective>,\n): Promise<Objective | null> {\n try {\n const existing = await sqlite.execute(\n `SELECT id, title, description, quarter, workspace, owner, status, progress,\n created_at, updated_at\n FROM objectives WHERE id = ?`,\n [id],\n );\n if (!existing.rows?.length) return null;\n const cur = rowToObjective(existing.columns, existing.rows[0] as any[]);\n const now = new Date().toISOString();\n const next = {\n title: data.title ?? cur.title,\n description: \"description\" in data ? (data.description ?? null) : cur.description,\n quarter: \"quarter\" in data ? (data.quarter ?? null) : cur.quarter,\n workspace: \"workspace\" in data ? (data.workspace ?? null) : cur.workspace,\n owner: \"owner\" in data ? (data.owner ?? null) : cur.owner,\n status: data.status ?? cur.status,\n };\n await sqlite.execute(\n `UPDATE objectives\n SET title = ?, description = ?, quarter = ?, workspace = ?, owner = ?,\n status = ?, updated_at = ?\n WHERE id = ?`,\n [next.title, next.description, next.quarter, next.workspace, next.owner, next.status, now, id],\n );\n const list = await listObjectives({});\n return list.find((o) => o.id === id) ?? null;\n } catch {\n return null;\n }\n}\n\nexport async function deleteObjective(id: string): Promise<boolean> {\n try {\n const check = await sqlite.execute(\"SELECT id FROM objectives WHERE id = ?\", [id]);\n if (!check.rows?.length) return false;\n // Manual cascade in case FK enforcement is off.\n await sqlite.execute(\"DELETE FROM key_results WHERE objective_id = ?\", [id]);\n await sqlite.execute(\"DELETE FROM objectives WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── Write (Key Results) ─────────────────────────────────────────────────────\n\nexport async function createKeyResult(\n objective_id: string,\n data: Partial<KeyResult>,\n): Promise<KeyResult> {\n const now = new Date().toISOString();\n const start_value = Number(data.start_value) || 0;\n const current_value = Number(data.current_value ?? data.start_value ?? 0);\n const target_value = Number(data.target_value) || 0;\n // QA fix: auto-normalize confidence — chat tools + UI sliders may send 1-10 instead of\n // the schema's 0-1. Detect any value > 1 and divide by 10.\n const rawConf = Number(data.confidence ?? 0.5);\n const confidence = clamp(rawConf > 1 ? rawConf / 10 : rawConf, 0, 1);\n\n const initial: KeyResult = {\n id: \"0\", // placeholder, replaced after insert\n objective_id,\n title: data.title ?? \"Untitled KR\",\n metric_source: data.metric_source ?? null,\n metric_key: data.metric_key ?? null,\n start_value,\n current_value,\n target_value,\n unit: data.unit ?? null,\n confidence,\n progress: 0,\n updated_at: now,\n };\n initial.progress = computeKRProgress(initial);\n\n await sqlite.execute(\n `INSERT INTO key_results\n (objective_id, title, metric_source, metric_key, start_value, current_value,\n target_value, unit, confidence, progress, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n initial.objective_id, initial.title, initial.metric_source,\n initial.metric_key, initial.start_value, initial.current_value,\n initial.target_value, initial.unit, initial.confidence, initial.progress, now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n initial.id = String((rowIdResult.rows[0] as any[])[0]);\n return initial;\n}\n\nexport async function updateKeyResult(\n id: string,\n data: Partial<KeyResult>,\n): Promise<KeyResult | null> {\n try {\n const existing = await sqlite.execute(\n `SELECT id, objective_id, title, metric_source, metric_key, start_value,\n current_value, target_value, unit, confidence, progress, updated_at\n FROM key_results WHERE id = ?`,\n [id],\n );\n if (!existing.rows?.length) return null;\n const cur = rowToKR(existing.columns, existing.rows[0] as any[]);\n const next: KeyResult = {\n ...cur,\n title: data.title ?? cur.title,\n metric_source: \"metric_source\" in data ? (data.metric_source ?? null) : cur.metric_source,\n metric_key: \"metric_key\" in data ? (data.metric_key ?? null) : cur.metric_key,\n start_value: data.start_value != null ? Number(data.start_value) : cur.start_value,\n current_value: data.current_value != null ? Number(data.current_value) : cur.current_value,\n target_value: data.target_value != null ? Number(data.target_value) : cur.target_value,\n unit: \"unit\" in data ? (data.unit ?? null) : cur.unit,\n confidence: data.confidence != null\n ? clamp((() => { const c = Number(data.confidence); return c > 1 ? c / 10 : c; })(), 0, 1)\n : cur.confidence,\n updated_at: new Date().toISOString(),\n };\n next.progress = computeKRProgress(next);\n await sqlite.execute(\n `UPDATE key_results\n SET title = ?, metric_source = ?, metric_key = ?, start_value = ?,\n current_value = ?, target_value = ?, unit = ?, confidence = ?,\n progress = ?, updated_at = ?\n WHERE id = ?`,\n [\n next.title, next.metric_source, next.metric_key, next.start_value,\n next.current_value, next.target_value, next.unit, next.confidence,\n next.progress, next.updated_at, id,\n ],\n );\n return next;\n } catch {\n return null;\n }\n}\n\n// ─── Dashboard ───────────────────────────────────────────────────────────────\n\nexport async function getOKRDashboard(): Promise<{\n total: number;\n on_track: number;\n at_risk: number;\n avg_progress: number;\n quarter: string | null;\n}> {\n const fallback = { total: 0, on_track: 0, at_risk: 0, avg_progress: 0, quarter: null as string | null };\n try {\n const objectives = await listObjectives({});\n if (!objectives.length) return fallback;\n\n let on_track = 0;\n let at_risk = 0;\n let progressSum = 0;\n let progressCount = 0;\n for (const o of objectives) {\n // Aggregate KR confidence per objective so we can compare progress vs\n // expected progress at that confidence level.\n const avgConfidence = o.key_results.length\n ? o.key_results.reduce((a, k) => a + Number(k.confidence || 0), 0) /\n o.key_results.length\n : 0.5;\n const expected = avgConfidence * 100;\n if (o.progress >= expected) on_track += 1;\n if (o.progress < expected - 15) at_risk += 1;\n progressSum += o.progress;\n progressCount += 1;\n }\n const avg_progress = progressCount > 0\n ? Math.round(progressSum / progressCount)\n : 0;\n\n // Pick the most common quarter as the current quarter label.\n const counts = new Map<string, number>();\n for (const o of objectives) {\n if (!o.quarter) continue;\n counts.set(o.quarter, (counts.get(o.quarter) || 0) + 1);\n }\n let quarter: string | null = null;\n let max = 0;\n for (const [q, n] of counts) {\n if (n > max) { max = n; quarter = q; }\n }\n // QA fix: when there are no objectives, fall back to the CURRENT quarter computed\n // from today's date rather than returning null (which the UI was rendering as \"—\").\n if (!quarter) {\n const now = new Date();\n const q = Math.floor(now.getUTCMonth() / 3) + 1;\n quarter = `Q${q}-${now.getUTCFullYear()}`;\n }\n\n return { total: objectives.length, on_track, at_risk, avg_progress, quarter };\n } catch {\n return { ...fallback, quarter: (() => {\n const now = new Date();\n const q = Math.floor(now.getUTCMonth() / 3) + 1;\n return `Q${q}-${now.getUTCFullYear()}`;\n })() };\n }\n}\n",[],[365,746],"29b479bc-5d69-11f1-98bb-1607ee4eb77e","nsm.ts","964e9fd2-589e-11f1-bbad-ee650bb23af1",620,["D",1780284167768],"// North Star Metric — Phase 1 Framework Layer.\n// Singleton NSM row (one per workspace at a time) + rolling history for trend\n// + velocity. Auto-pulls live value from the funnel summary when the NSM is\n// pinned to a connected integration metric (e.g. ga4:conversions).\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp = (n: number, lo: number, hi: number) =>\n Math.max(lo, Math.min(hi, n));\n\nconst daysBetween = (laterISO: string, earlierISO: string) =>\n (new Date(laterISO).getTime() - new Date(earlierISO).getTime()) / 86400000;\n\nexport interface NSMRow {\n id: string;\n metric_name: string;\n metric_source: string | null;\n metric_key: string | null;\n current_value: number;\n target_value: number;\n target_date: string | null;\n unit: string | null;\n description: string | null;\n set_at: string;\n set_by: string | null;\n}\n\n// Resolve a live metric value from the unified funnel summary when the NSM is\n// pinned to a known integration metric. Returns null when there's no match\n// (e.g. for a manually-tracked north star like \"revenue\" with no integration).\nasync function resolveLiveValue(\n metric_source: string | null,\n metric_key: string | null,\n): Promise<number | null> {\n if (!metric_source || !metric_key) return null;\n try {\n const summary = await getFunnelSummary();\n const metrics = summary.stages.flatMap((s) => s.metrics);\n const hit = metrics.find(\n (m: any) => m.integration === metric_source && m.metric === metric_key,\n );\n if (hit && typeof (hit as any).value === \"number\") {\n return Number((hit as any).value);\n }\n return null;\n } catch {\n return null;\n }\n}\n\n// ─── CRUD ────────────────────────────────────────────────────────────────────\n\nexport async function getNSM(): Promise<\n { nsm: NSMRow | null; history: Array<{ value: number; recorded_at: string }> }\n> {\n try {\n const r = await sqlite.execute(\n `SELECT id, metric_name, metric_source, metric_key, current_value, target_value,\n target_date, unit, description, set_at, set_by\n FROM north_star\n ORDER BY set_at DESC\n LIMIT 1`,\n );\n if (!r.rows?.length) return { nsm: null, history: [] };\n const obj = rowToObj(r.columns, r.rows[0] as any[]);\n\n // Sync live value if pinned to a connected integration.\n const live = await resolveLiveValue(obj.metric_source, obj.metric_key);\n if (live !== null && Number.isFinite(live) && live !== Number(obj.current_value)) {\n try {\n await sqlite.execute(\n \"UPDATE north_star SET current_value = ? WHERE id = ?\",\n [live, obj.id],\n );\n obj.current_value = live;\n // Append to history so the cron/dashboard sees the refreshed point.\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, datetime('now'))\",\n [obj.id, live],\n );\n } catch { /* non-critical */ }\n }\n\n const nsm: NSMRow = {\n id: String(obj.id),\n metric_name: String(obj.metric_name ?? \"\"),\n metric_source: obj.metric_source ?? null,\n metric_key: obj.metric_key ?? null,\n current_value: Number(obj.current_value) || 0,\n target_value: Number(obj.target_value) || 0,\n target_date: obj.target_date ?? null,\n unit: obj.unit ?? null,\n description: obj.description ?? null,\n set_at: String(obj.set_at ?? new Date().toISOString()),\n set_by: obj.set_by ?? null,\n };\n\n const hist = await sqlite.execute(\n `SELECT value, recorded_at\n FROM nsm_history\n WHERE nsm_id = ?\n AND recorded_at >= datetime('now','-90 days')\n ORDER BY recorded_at ASC`,\n [nsm.id],\n );\n const history = (hist.rows || []).map((row: any) => ({\n value: Number(row[0]) || 0,\n recorded_at: String(row[1]),\n }));\n\n return { nsm, history };\n } catch {\n return { nsm: null, history: [] };\n }\n}\n\nexport async function setNSM(data: {\n metric_name: string;\n metric_source?: string | null;\n metric_key?: string | null;\n target_value: number;\n target_date?: string | null;\n unit?: string | null;\n description?: string | null;\n set_by?: string | null;\n}): Promise<NSMRow> {\n const now = new Date().toISOString();\n\n // Singleton semantics: clear out previous NSM(s) + their history.\n try {\n const prior = await sqlite.execute(\"SELECT id FROM north_star\");\n for (const row of (prior.rows || []) as any[]) {\n const priorId = String(row[0]);\n await sqlite.execute(\"DELETE FROM nsm_history WHERE nsm_id = ?\", [priorId]);\n }\n await sqlite.execute(\"DELETE FROM north_star\");\n } catch { /* non-critical */ }\n\n // Seed current_value from the live integration value when pinned, else 0.\n const live = await resolveLiveValue(\n data.metric_source ?? null,\n data.metric_key ?? null,\n );\n const current_value = live !== null && Number.isFinite(live) ? live : 0;\n const target_date = data.target_date ?? now.slice(0, 10);\n\n await sqlite.execute(\n `INSERT INTO north_star\n (metric_name, metric_source, metric_key, current_value, target_value,\n target_date, unit, description, set_at, set_by)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.metric_name,\n data.metric_source ?? \"\", // NOT NULL column — default to empty string for manual NSMs\n data.metric_key ?? null,\n current_value,\n Number(data.target_value) || 0,\n target_date,\n data.unit ?? null,\n data.description ?? null,\n now,\n data.set_by ?? null,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n\n // Seed nsm_history only when we already have a non-zero starting value (e.g. pulled live\n // from the connected integration). User QA flagged a 0-value blip when this fired before\n // the first metric sync — that blip dragged velocity calculations to -∞.\n if (current_value > 0) {\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, ?)\",\n [id, current_value, now],\n );\n }\n\n return {\n id,\n metric_name: data.metric_name,\n metric_source: data.metric_source ?? \"\",\n metric_key: data.metric_key ?? null,\n current_value,\n target_value: Number(data.target_value) || 0,\n target_date,\n unit: data.unit ?? null,\n description: data.description ?? null,\n set_at: now,\n set_by: data.set_by ?? null,\n };\n}\n\nexport async function recordNSMValue(value: number): Promise<void> {\n try {\n const r = await sqlite.execute(\n \"SELECT id FROM north_star ORDER BY set_at DESC LIMIT 1\",\n );\n if (!r.rows?.length) return;\n const nsmId = String((r.rows[0] as any[])[0]);\n await sqlite.execute(\n \"INSERT INTO nsm_history (nsm_id, value, recorded_at) VALUES (?, ?, datetime('now'))\",\n [nsmId, Number(value) || 0],\n );\n await sqlite.execute(\n \"UPDATE north_star SET current_value = ? WHERE id = ?\",\n [Number(value) || 0, nsmId],\n );\n } catch { /* non-critical */ }\n}\n\n// ─── Progress / Velocity ─────────────────────────────────────────────────────\n\nexport interface NSMProgress {\n current: number;\n target: number;\n target_date: string | null;\n pct_complete: number; // 0..100\n days_remaining: number | null; // until target_date\n on_track: boolean;\n velocity_per_day: number | null;\n days_to_target: number | null; // null when velocity ≤ 0 or no history\n}\n\nexport async function getNSMProgress(): Promise<NSMProgress> {\n const fallback: NSMProgress = {\n current: 0,\n target: 0,\n target_date: null,\n pct_complete: 0,\n days_remaining: null,\n on_track: false,\n velocity_per_day: null,\n days_to_target: null,\n };\n\n try {\n const { nsm, history } = await getNSM();\n if (!nsm) return fallback;\n\n const current = Number(nsm.current_value) || 0;\n const target = Number(nsm.target_value) || 0;\n\n // Velocity = (current - value_30d_ago) / 30. Use the oldest sample within\n // the trailing 30d window when an exact 30d-old point isn't present.\n let value30dAgo: number | null = null;\n if (history.length > 0) {\n const thirtyAgo = Date.now() - 30 * 86400000;\n // Earliest sample >= 30d ago, or fall back to the first sample we have.\n const inWindow = history.filter(\n (h) => new Date(h.recorded_at).getTime() <= thirtyAgo,\n );\n if (inWindow.length > 0) {\n value30dAgo = inWindow[inWindow.length - 1].value;\n } else {\n value30dAgo = history[0].value;\n }\n }\n\n const velocity_per_day =\n value30dAgo !== null ? (current - value30dAgo) / 30 : null;\n\n const days_to_target =\n velocity_per_day !== null && velocity_per_day > 0\n ? (target - current) / velocity_per_day\n : null;\n\n const days_remaining =\n nsm.target_date\n ? Math.max(0, daysBetween(nsm.target_date, new Date().toISOString()))\n : null;\n\n const on_track =\n days_to_target !== null &&\n days_remaining !== null &&\n days_to_target <= days_remaining;\n\n const pct_complete = target > 0 ? clamp((current / target) * 100, 0, 100) : 0;\n\n return {\n current,\n target,\n target_date: nsm.target_date,\n pct_complete,\n days_remaining,\n on_track,\n velocity_per_day,\n days_to_target,\n };\n } catch {\n return fallback;\n }\n}\n",[],[365,754],"6973f69c-5989-11f1-aff7-ee650bb23af1","method_config.ts","6973f750-5989-11f1-aff7-ee650bb23af1",366,["D",1779858213970],"// Configurable Metrics + Growth Methodology.\n// Backs the Settings → \"Growth Methodology\" panel. Lets users:\n// 1) Pick a prioritization method (ICE | RICE | WSJF) — drives suggestICE + experiment scoring.\n// 2) Override funnel-stage metrics — e.g. swap GSC clicks → impressions on TOFU.\n// 3) Define custom KPIs (name + formula + unit) that surface on Analytics.\n//\n// Storage uses the existing `app_config` (key/value JSON) table — no new schema.\nimport { sqlite } from \"../db.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type PrioritizationMethod = \"ICE\" | \"RICE\" | \"WSJF\";\n\nexport interface MetricOverride {\n integration: string; // e.g. \"gsc\", \"ga4\"\n metric: string; // e.g. \"impressions\", \"ctr\"\n weight: number; // 0..2 — multiplier applied during health computation\n}\n\nexport interface CustomKPI {\n name: string; // user-defined label\n formula: string; // free-form expression (eval'd in Analytics layer; stored as text)\n unit: string; // \"count\", \"%\", \"$\", \"s\", etc.\n}\n\nexport interface FunnelMetricOverrides {\n TOFU: MetricOverride[];\n MOFU: MetricOverride[];\n BOFU: MetricOverride[];\n}\n\nexport interface MethodConfig {\n prioritization_method: PrioritizationMethod;\n funnel_metric_overrides: FunnelMetricOverrides;\n custom_kpis: CustomKPI[];\n}\n\n// ─── Static method metadata ──────────────────────────────────────────────────\n// `compute` returns the prioritization score; identical contract regardless of method\n// so callers can do `METHODS[m].compute(...args)` without branching.\n\nexport const METHODS = {\n ICE: {\n name: \"ICE\",\n formula: \"(Impact + Confidence + Ease) / 3\",\n range: \"0-10 each\",\n fields: [\"impact\", \"confidence\", \"ease\"] as string[],\n defaults: { impact: 5, confidence: 5, ease: 5 },\n compute: (i: number, c: number, e: number): number =>\n Math.round(((i + c + e) / 3) * 100) / 100,\n why:\n \"Fast triage. Best when you have many ideas and need to ship quickly. \" +\n \"Used by Sean Ellis at LogMeIn/Dropbox.\",\n },\n RICE: {\n name: \"RICE\",\n formula: \"(Reach × Impact × Confidence) / Effort\",\n range:\n \"Reach: # users/period, Impact: 0.25-3, Confidence: 0-100%, Effort: person-weeks\",\n fields: [\"reach\", \"impact\", \"confidence\", \"effort\"] as string[],\n defaults: { reach: 1000, impact: 1, confidence: 0.8, effort: 2 },\n compute: (r: number, i: number, c: number, e: number): number =>\n Math.round(((r * i * c) / Math.max(e, 0.1)) * 100) / 100,\n why:\n \"Better when reach varies a lot across experiments. Used by Intercom.\",\n },\n WSJF: {\n name: \"WSJF (Weighted Shortest Job First)\",\n formula: \"(Business Value + Time Criticality + Risk Reduction) / Job Size\",\n range: \"1-10 each (modified Fibonacci)\",\n fields: [\"business_value\", \"time_criticality\", \"risk_reduction\", \"job_size\"] as string[],\n defaults: { business_value: 5, time_criticality: 5, risk_reduction: 5, job_size: 3 },\n compute: (bv: number, tc: number, rr: number, js: number): number =>\n Math.round(((bv + tc + rr) / Math.max(js, 1)) * 100) / 100,\n why:\n \"SAFe / agile teams. Best when ladder of work item sizes matters.\",\n },\n} as const;\n\nexport type MethodKey = keyof typeof METHODS;\n\n// ─── Persistence helpers ─────────────────────────────────────────────────────\n\nconst KEY_METHOD = \"prioritization_method\";\nconst KEY_OVERRIDES = \"funnel_metric_overrides\";\nconst KEY_KPIS = \"custom_kpis\";\n\nasync function getConfigValue(key: string): Promise<string | null> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [key]);\n if (!r.rows?.length) return null;\n const v = (r.rows[0] as any[])[0];\n return v == null ? null : String(v);\n } catch {\n return null;\n }\n}\n\nasync function setConfigValue(key: string, value: string): Promise<void> {\n const now = new Date().toISOString();\n try {\n await sqlite.execute(\n \"INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, ?) \" +\n \"ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = excluded.updated_at\",\n [key, value, now],\n );\n } catch {\n // Fallback for sqlite builds without UPSERT — manual update-or-insert.\n try {\n const existing = await sqlite.execute(\"SELECT key FROM app_config WHERE key = ?\", [key]);\n if (existing.rows?.length) {\n await sqlite.execute(\"UPDATE app_config SET value = ?, updated_at = ? WHERE key = ?\", [value, now, key]);\n } else {\n await sqlite.execute(\"INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, ?)\", [key, value, now]);\n }\n } catch { /* swallow — non-critical */ }\n }\n}\n\nfunction normalizeMethod(raw: any): PrioritizationMethod {\n const s = String(raw ?? \"\").toUpperCase();\n if (s === \"RICE\" || s === \"WSJF\" || s === \"ICE\") return s as PrioritizationMethod;\n return \"ICE\";\n}\n\nfunction normalizeOverrides(raw: any): FunnelMetricOverrides {\n const empty: FunnelMetricOverrides = { TOFU: [], MOFU: [], BOFU: [] };\n if (!raw || typeof raw !== \"object\") return empty;\n const out: FunnelMetricOverrides = { TOFU: [], MOFU: [], BOFU: [] };\n for (const stage of [\"TOFU\", \"MOFU\", \"BOFU\"] as const) {\n const arr = Array.isArray(raw[stage]) ? raw[stage] : [];\n out[stage] = arr\n .filter((r: any) => r && typeof r === \"object\" && r.integration && r.metric)\n .map((r: any) => ({\n integration: String(r.integration),\n metric: String(r.metric),\n weight: Number.isFinite(Number(r.weight)) ? Number(r.weight) : 1,\n }));\n }\n return out;\n}\n\nfunction normalizeKpis(raw: any): CustomKPI[] {\n if (!Array.isArray(raw)) return [];\n return raw\n .filter((k: any) => k && typeof k === \"object\" && k.name && k.formula)\n .map((k: any) => ({\n name: String(k.name).slice(0, 80),\n formula: String(k.formula).slice(0, 500),\n unit: String(k.unit ?? \"count\").slice(0, 20),\n }));\n}\n\n// ─── Public API ──────────────────────────────────────────────────────────────\n\nexport async function getMethodConfig(): Promise<MethodConfig> {\n const [m, o, k] = await Promise.all([\n getConfigValue(KEY_METHOD),\n getConfigValue(KEY_OVERRIDES),\n getConfigValue(KEY_KPIS),\n ]);\n let overridesRaw: any = null;\n let kpisRaw: any = null;\n try { if (o) overridesRaw = JSON.parse(o); } catch { /* malformed */ }\n try { if (k) kpisRaw = JSON.parse(k); } catch { /* malformed */ }\n return {\n prioritization_method: normalizeMethod(m),\n funnel_metric_overrides: normalizeOverrides(overridesRaw),\n custom_kpis: normalizeKpis(kpisRaw),\n };\n}\n\nexport async function saveMethodConfig(patch: Partial<MethodConfig>): Promise<MethodConfig> {\n const current = await getMethodConfig();\n const next: MethodConfig = {\n prioritization_method: patch.prioritization_method !== undefined\n ? normalizeMethod(patch.prioritization_method)\n : current.prioritization_method,\n funnel_metric_overrides: patch.funnel_metric_overrides !== undefined\n ? normalizeOverrides(patch.funnel_metric_overrides)\n : current.funnel_metric_overrides,\n custom_kpis: patch.custom_kpis !== undefined\n ? normalizeKpis(patch.custom_kpis)\n : current.custom_kpis,\n };\n await Promise.all([\n setConfigValue(KEY_METHOD, next.prioritization_method),\n setConfigValue(KEY_OVERRIDES, JSON.stringify(next.funnel_metric_overrides)),\n setConfigValue(KEY_KPIS, JSON.stringify(next.custom_kpis)),\n ]);\n return next;\n}\n\n// Helper used by api/experiments.ts to fetch JUST the method (cheaper, no JSON parse).\nexport async function getPrioritizationMethod(): Promise<PrioritizationMethod> {\n return normalizeMethod(await getConfigValue(KEY_METHOD));\n}\n\n// Static metadata payload for GET /api/method-config/methods — surfaces formula + why-to-use\n// in the Settings UI without round-trips. Strips the `compute` function (non-serializable).\nexport function getMethodsMetadata(): Array<{\n key: MethodKey;\n name: string;\n formula: string;\n range: string;\n fields: string[];\n defaults: Record<string, number>;\n why: string;\n}> {\n return (Object.keys(METHODS) as MethodKey[]).map((key) => {\n const m = METHODS[key];\n return {\n key,\n name: m.name,\n formula: m.formula,\n range: m.range,\n fields: [...m.fields],\n defaults: { ...m.defaults },\n why: m.why,\n };\n });\n}\n",[],[365,762],"d48917c4-5d61-11f1-9b76-1607ee4eb77e","memory.ts","d489186e-5d61-11f1-9b76-1607ee4eb77e",578,["D",1780281018405],"// Part 15.3 — Memory: persistent, evolving business memory.\n// Company Profile = static identity; Memory = evolving learned context the AI accumulates and\n// reuses across conversations. CRUD + retrieval + a compact injection string + a best-effort\n// auto-capture that distills durable facts from a finished conversation.\nimport { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\nexport interface MemoryItem {\n id: number;\n type: string; // company | preference | fact | decision | learning\n content: string;\n source: string; // chat | agent | experiment | manual\n confidence: number;\n pinned: boolean;\n created_at: string;\n updated_at: string;\n}\n\nconst VALID_TYPES = new Set([\"company\", \"preference\", \"fact\", \"decision\", \"learning\"]);\n\nfunction rowTo(cols: string[], row: any[]): MemoryItem {\n const o: any = {};\n for (let i = 0; i < cols.length; i++) o[cols[i]] = row[i];\n return { ...o, pinned: !!o.pinned, confidence: Number(o.confidence) };\n}\n\nexport async function listMemory(): Promise<MemoryItem[]> {\n try {\n const r = await sqlite.execute(\n \"SELECT id, type, content, source, confidence, pinned, created_at, updated_at FROM memory ORDER BY pinned DESC, updated_at DESC LIMIT 200\"\n );\n return (r.rows || []).map((row) => rowTo(r.columns, row as any[]));\n } catch { return []; }\n}\n\nfunction normContent(s: string): string {\n return s.trim().toLowerCase().replace(/\\s+/g, \" \");\n}\n\nexport async function addMemory(data: { content: string; type?: string; source?: string; confidence?: number; pinned?: boolean }): Promise<MemoryItem | null> {\n const content = String(data.content || \"\").trim();\n if (!content) return null;\n const type = data.type && VALID_TYPES.has(data.type) ? data.type : \"fact\";\n const source = (data.source || \"manual\").toString();\n const confidence = typeof data.confidence === \"number\" ? Math.max(0, Math.min(1, data.confidence)) : 0.7;\n const pinned = data.pinned ? 1 : 0;\n\n // Dedupe: if an existing memory has the same normalized content, bump it instead of duplicating.\n try {\n const existing = await sqlite.execute(\"SELECT id, content FROM memory\");\n for (const row of (existing.rows || []) as any[]) {\n if (normContent(String(row[1])) === normContent(content)) {\n await sqlite.execute(\"UPDATE memory SET updated_at = datetime('now'), confidence = ? WHERE id = ?\", [Math.max(confidence, 0.7), row[0]]);\n return await getMemory(Number(row[0]));\n }\n }\n } catch { /* table optional */ }\n\n await sqlite.execute(\n \"INSERT INTO memory (type, content, source, confidence, pinned) VALUES (?, ?, ?, ?, ?)\",\n [type, content, source, confidence, pinned]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n return await getMemory(Number((idRow.rows[0] as any[])[0]));\n}\n\nexport async function getMemory(id: number): Promise<MemoryItem | null> {\n const r = await sqlite.execute(\n \"SELECT id, type, content, source, confidence, pinned, created_at, updated_at FROM memory WHERE id = ?\", [id]\n );\n if (!r.rows?.length) return null;\n return rowTo(r.columns, r.rows[0] as any[]);\n}\n\nexport async function updateMemory(id: number, data: { content?: string; type?: string; pinned?: boolean; confidence?: number }): Promise<MemoryItem | null> {\n const cur = await getMemory(id);\n if (!cur) return null;\n const content = typeof data.content === \"string\" ? data.content.trim() : cur.content;\n const type = data.type && VALID_TYPES.has(data.type) ? data.type : cur.type;\n const pinned = typeof data.pinned === \"boolean\" ? (data.pinned ? 1 : 0) : (cur.pinned ? 1 : 0);\n const confidence = typeof data.confidence === \"number\" ? Math.max(0, Math.min(1, data.confidence)) : cur.confidence;\n await sqlite.execute(\n \"UPDATE memory SET content = ?, type = ?, pinned = ?, confidence = ?, updated_at = datetime('now') WHERE id = ?\",\n [content, type, pinned, confidence, id]\n );\n return await getMemory(id);\n}\n\nexport async function deleteMemory(id: number): Promise<boolean> {\n try { await sqlite.execute(\"DELETE FROM memory WHERE id = ?\", [id]); return true; } catch { return false; }\n}\n\n// Compact injection: pinned first, then most-recent, capped (~2KB per Part 15.4 budget).\nexport async function getMemoryContextString(maxChars = 2000): Promise<string> {\n const items = await listMemory();\n if (!items.length) return \"\";\n let out = \"## Memory (what we've learned about this business)\\n\";\n let remaining = maxChars - out.length;\n for (const m of items) {\n const line = `- [${m.type}] ${m.content}\\n`;\n if (line.length > remaining) break;\n out += line;\n remaining -= line.length;\n }\n return out.trim();\n}\n\n// Best-effort auto-capture: distill durable facts from a finished conversation. Fire-and-forget;\n// never block the chat response. Only runs when an LLM key exists and there's enough signal.\nexport async function captureMemoryFromConversation(messages: Array<{ role: string; content: string }>): Promise<void> {\n try {\n const convo = messages.filter((m) => m.role === \"user\" || m.role === \"assistant\")\n .map((m) => `${m.role}: ${String(m.content || \"\").slice(0, 1200)}`).join(\"\\n\");\n if (convo.length < 200) return; // too little to learn from\n const cfg = await getLLMConfig();\n if (!(cfg as any).api_key) return;\n\n const prompt = `From the conversation below, extract 0-3 DURABLE facts worth remembering about the\nuser's BUSINESS (not transient task details). Examples of durable: target market/ICP, tone preference,\na stated goal/target, a strategic decision, a learned constraint. Ignore one-off questions and pleasantries.\nReturn ONLY a JSON array of objects: [{\"type\":\"company|preference|fact|decision|learning\",\"content\":\"<concise fact>\"}].\nReturn [] if nothing is durable.\n\nCONVERSATION:\n${convo.slice(0, 6000)}`;\n\n const resp = await callLLM(cfg, [{ role: \"user\", content: prompt }], { maxTokens: 400 });\n const text = (resp.content || \"\").trim();\n const match = text.match(/\\[[\\s\\S]*\\]/);\n if (!match) return;\n let arr: any[] = [];\n try { arr = JSON.parse(match[0]); } catch { return; }\n for (const item of arr.slice(0, 3)) {\n if (item && typeof item.content === \"string\" && item.content.trim()) {\n await addMemory({ content: item.content, type: item.type, source: \"chat\", confidence: 0.6 });\n }\n }\n } catch { /* non-fatal */ }\n}\n",[],[365,770],"bdb7148e-5a4f-11f1-9df7-ee650bb23af1","math_knowledge_seed.ts","3acb1cf0-5991-11f1-8131-ee650bb23af1",404,["D",1779943395684],"// Math/framework knowledge seeds for Nexus.\n// Loaded into knowledge_files on init. LLM prompts inject\n// rows whose `targets` match the active feature. Empty = global.\n\nexport interface MathKnowledgeSeed {\n name: string; // human label\n kind: 'formula' | 'framework' | 'benchmark';\n targets: string[]; // which AI features consume this\n content: string; // 400-800 words of dense, AI-readable markdown\n}\n\nexport const MATH_KNOWLEDGE_SEEDS: MathKnowledgeSeed[] = [\n {\n name: 'Customer Acquisition Cost (CAC)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest'],\n content: `Source: SaaS unit economics (David Skok, Bessemer State of the Cloud).\n\n### Formula\n- CAC = est_marketing_spend / new_customers_acquired\n- Window: trailing 30 or 90 days. Spend window MUST match customer window.\n- Fully-loaded CAC = (marketing + sales_payroll + tooling + agency) / new_customers. Nexus uses paid-channel spend only — state this caveat when comparing to public benchmarks.\n- Blended CAC = total_spend / total_new_customers. Channel CAC = channel_spend / channel_attributed_customers.\n\n### Thresholds\n- Payback period = CAC / (ARPU × gross_margin). Healthy SaaS: < 12 months. World-class: < 6 months.\n- CAC must be < LTV / 3. If CAC ≥ LTV/3 → UNSUSTAINABLE.\n- B2B SaaS medians: $700 SMB, $7k mid-market, $25k+ enterprise (OpenView 2024).\n- Ecommerce: blended CAC ≤ 30% of first-order AOV for repeat categories.\n\n### When/Then\n- CAC trend rises > 20% MoM AND new_customers flat → channel saturation / creative fatigue. Action: rotate creative, test new channel.\n- CAC rises and LTV rises proportionally → healthy if LTV/CAC ≥ 3 holds.\n- CAC > LTV/3 → flag UNSUSTAINABLE in digest. Recommend pausing worst-MER channel.\n- Channel CAC < blended × 0.7 → scale that channel.\n- Channel CAC > blended × 1.5 → cut or restructure.\n- Rising CAC + falling MER + flat conversion = paid inefficiency (creative/audience).\n- Rising CAC + falling conversion = funnel quality issue (landing page / ICP).\n\n### Anti-pattern\n- Comparing CAC across different window lengths — always normalize.\n- Excluding organic customers from denominator while including all spend → inflated CAC.\n- Ignoring sales-cycle lag: customers this month caused by spend ~lag days ago. Lag-correct when sales_cycle > window.\n\n### Interpretation for chat\n- \"Is our CAC OK?\" → answer with current CAC, LTV/CAC ratio, payback period. Verdict only if all three computable.\n- If CAC missing, do not fabricate. State which input is missing and which integration supplies it (Stripe/HubSpot for customers, Meta/Google/LinkedIn for spend).\n- Rising trend: 3-week rolling vs prior 3-week. > 15% = \"rising,\" > 30% = \"spiking.\" Cross-check seasonality / campaign launches first.\n\n### Decision rules (Nexus)\n- payback > 18mo → digest \"Cash-efficiency risk.\"\n- LTV/CAC < 3 → recommendations: \"Tighten ICP\" or \"Cut worst channel.\"\n- CAC trend down + LTV trend up → digest \"Compounding efficiency.\"\n- Channel reallocation only if delta vs blended > 30% AND channel spend > $1k/week (noise filter).\n- One paid channel → blended only; recommend channel diversification.\n- CAC from < 14d data → label \"preliminary\"; require ≥ 30d before automated alerts.`,\n },\n\n {\n name: 'Lifetime Value (LTV)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest', 'cohort_ltv'],\n content: `Source: David Skok (For Entrepreneurs), Bessemer cohort LTV.\n\n### Formula\n- Simple LTV = avg_deal_size × retention_months\n- Margin-adjusted LTV = ARPU × gross_margin × (1 / monthly_churn)\n- Cohort survival LTV = ARPU × Σ retention_rate_t for t=1..N. Sum the observed cohort survival curve; do NOT extrapolate beyond observed data unless fitting a curve (geometric / Kaplan-Meier).\n- Discounted LTV = Σ(ARPU_t × retention_t / (1+r)^t).\n\n### Thresholds\n- LTV/CAC ≥ 3 = healthy. ≥ 5 = world-class (Bessemer top quartile). < 1 = burning cash per customer.\n- NRR benchmark: ≥ 100% (best-in-class ≥ 120%).\n- GRR benchmark: ≥ 90% SMB, ≥ 95% mid+.\n- Monthly logo churn: < 1% excellent, 1-3% acceptable SMB, > 5% critical.\n\n### When/Then\n- LTV flat 3 months AND CAC rising → CAC efficiency degrading. Focus on retention/expansion before more acquisition.\n- LTV rises via expansion (NRR > 100%) → double acquisition spend; payback shortens.\n- LTV rises but only top-decile customers expand → median LTV may be falling. Segment-level analysis required.\n- Cohort survival curves diverge by channel → channel quality differs. Allocate to high-LTV channels even at higher CAC.\n- Monthly churn > 5% → cap LTV reporting; LTV math unstable (sensitive to small churn shifts).\n\n### Cohort survival LTV (api/aio.ts)\n- For each cohort C_m, retention_t = active(m+t) / cohort_size.\n- LTV(C_m, N) = ARPU(C_m) × Σ retention_t for t=0..N. Default N=24mo.\n- Report curve shape: \"flattening after month 6\" = sticky, \"linear decay\" = transactional.\n- Cohorts < 6 months: label \"preliminary\"; use blended curve as prior.\n\n### Anti-pattern\n- 1/churn with < 12 months data — extrapolation explodes (1/1% = 100 months projected).\n- Mixing gross/net revenue — pick one.\n- Ignoring gross margin. $100 ARPU at 30% margin → effective $30 LTV/mo.\n- Single blended LTV across SMB+Enterprise — segment when divergence > 2x.\n\n### Interpretation for chat\n- \"What's our LTV?\" → return ARPU, gross margin, churn, resulting LTV. State which formula (simple / cohort / margin-adjusted).\n- LTV is a forecast, not a fact. Communicate uncertainty: cohort age + sample size.\n- Refuse single LTV number when segments diverge > 2x — break it down.\n\n### Decision rules\n- LTV/CAC < 3 + churn > 5% → recommendations: \"Retention sprint\" over acquisition.\n- LTV trend up 10%+ / 90d AND CAC flat → digest \"Compounding.\"\n- Cohort curve flattens by month 3 → PMF signal; recommend acquisition scale-up.`,\n },\n\n {\n name: 'Marketing Efficiency Ratio (MER)',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest'],\n content: `Source: Triple Whale / Klaviyo / Common Thread Collective MER convention.\n\n### Formula\n- MER = total_revenue / total_marketing_spend\n- Blended MER = ALL revenue (organic + paid + direct) over ALL paid spend. Headline number.\n- Channel MER = channel_attributed_revenue / channel_spend. Use last-click for simplicity.\n- nMER (new-customer MER) = new_customer_revenue / spend. More honest for acquisition.\n- aMER = first-order_revenue_from_new / spend.\n\n### Thresholds\n- Blended MER ≥ 3.0x = healthy. ≥ 5.0x = world-class. < 2.0x = unprofitable for most categories.\n- nMER ≥ 1.5x = acceptable at scale. ≥ 2.5x = strong.\n- Platform ROAS typically inflates by 2-3x vs MER. Trust MER for budget decisions.\n- Category: DTC apparel 2.5-3.5x, supplements 3.5-5x, B2B SaaS 4-8x.\n\n### When/Then\n- Channel MER < blended → dragging average down. Divest or restructure.\n- Channel MER > blended by > 50% → scale +20%, re-measure after 7 days.\n- Blended MER falls > 20% WoW → diagnose: spend up? revenue down? Both? Each implies different action.\n- nMER < 1 but blended > 3 → acquisition unprofitable; base carrying. Short-term only.\n- MER rises but new-customer count falls → harvesting existing demand. Collapses when base saturates.\n\n### Channel decision rule\nFor each channel with spend > $1k/week:\n- Score = (MER_c - blended) / blended\n- > +0.3 → +20% budget. < -0.2 → -50% or pause. Between → hold.\n\n### Anti-pattern\n- Quoting platform ROAS (Meta says 4x) as if equivalent to blended MER (1.8x). Platform ROAS double-counts.\n- 7-day MER window when sales cycle is 30d → lag artifacts. Use cycle-aligned window.\n- Excluding offline/email revenue → MER inflates (spend includes channels driving offline).\n- Counting subscription renewals in MER → renewals not caused by current spend. Use nMER for acquisition.\n\n### Interpretation for chat\n- \"Is paid working?\" → quote blended MER, nMER, and single best-delta channel. Always pair MER with trend.\n- MER alone insufficient. Pair with CAC trend + customer count to avoid the harvesting trap.\n- Use MER, not ROAS, in any cross-channel comparison.\n\n### Decision rules\n- Blended MER < 2.0x for 2 weeks → recommendations: \"Spend efficiency review.\"\n- Channel MER < 1.0x for 1 week → digest \"Loss-making channel: <name>.\"\n- nMER < 1.0x with blended ≥ 3.0x → digest \"Acquisition unprofitable; base carrying.\"\n- MER drop of > 30% week-over-week → standup headline regardless of absolute level.\n- For attribution-incremental teams: pair MER with a quarterly geo-lift or holdout test; MER alone over-credits brand/organic compounding to paid channels.`,\n },\n\n {\n name: 'Pipeline Coverage Ratio',\n kind: 'formula',\n targets: ['kpis', 'chat', 'digest', 'standup'],\n content: `Source: Sales-ops standard (Salesforce State of Sales, RevOps Co-op).\n\n### Formula\n- Coverage = open_pipeline_value / sales_quota_remaining (same period)\n- Pipeline = Σ deal_amount for stages discovery → proposal → negotiation. Exclude closed-won/lost.\n- Weighted coverage = Σ(deal × stage_probability) / quota. Weights: discovery 10%, qualified 25%, proposal 50%, negotiation 75%.\n- Forward coverage = next-quarter open pipeline / next-quarter quota. Early warning.\n\n### Thresholds\n- Healthy: ≥ 3.0x raw for next quarter (assumes ~33% win rate).\n- At-risk: 2.0-2.9x — likely to miss without intervention.\n- Off-track: < 2.0x — will miss barring outliers.\n- World-class: 4-5x at Q start, declining to 1.5-2x at Q end.\n- Win-rate-adjusted target = 1 / win_rate. 25% win rate → need 4x. 40% → 2.5x.\n\n### When/Then\n- Coverage < 2.0x with > 6 weeks left → trigger ToFu campaign (paid + outbound).\n- Coverage < 2.0x with < 4 weeks left → triage existing; chase commit-stage only. New pipeline won't close.\n- Coverage > 5x AND win rate trending down → bloated with low-quality deals. Pipeline hygiene sweep.\n- Coverage stable but avg deal size falling → ICP drift; review ICP.\n- Forward coverage (next Q) < 1.5x at current Q midpoint → escalate; pipeline build too slow.\n\n### Stage velocity check\n- Track median days_in_stage. If rises 30%+ vs trailing 90d → deals stuck. Coverage may be inflated by zombies.\n- Auto-flag deals with days_in_stage > 2× median for AE review.\n\n### Diagnostic decomposition\nShortfall = (quota - covered) / avg_deal_size = N new deals needed.\n- N > 10× current weekly new-deal rate → unrealistic; cut quota.\n- N achievable → recommend specific outbound sequences / paid bump.\n\n### Anti-pattern\n- Counting same deal across periods (rolling 12-mo pipeline vs quarterly quota) → inflates 3-4x.\n- Ignoring win-rate drift. Static 3x target wrong if win rate dropped 30%→20% — need 5x.\n- Treating all pipeline equally. Late-stage worth ~5x early-stage in real revenue.\n- Pumping low-quality deals to hit coverage → destroys win rate + forecast accuracy.\n- Unweighted as only number. Weighted = truer signal.\n\n### Interpretation for chat\n- \"Are we on track this quarter?\" → return weighted coverage, raw coverage, win rate, deals-needed-to-hit. All four.\n- Coverage requires quota. If unknown, ask once, then compute.\n- Pair coverage with stage-velocity sanity check; high coverage + slow velocity = false comfort.\n\n### Decision rules (Nexus)\n- Coverage < 2.0x → standup headline \"Pipeline gap: need $X more open by EOQ.\"\n- Coverage 2.0-2.9x → digest \"At-risk; ToFu push recommended.\"\n- Coverage trending down 2 weeks → recommendations: \"Outbound sequence boost\" + \"Paid ToFu spike.\"\n- Coverage > 5x + stage velocity down 30% → recommendations: \"Pipeline hygiene sweep.\"`,\n },\n\n {\n name: 'ICE Prioritization (Sean Ellis)',\n kind: 'framework',\n targets: ['ice_suggest', 'chat'],\n content: `Source: Sean Ellis, 2010 (LogMeIn, Dropbox growth team). For high-velocity experiment triage.\n\n### Formula\n- ICE = (Impact + Confidence + Ease) / 3\n- Each axis 1-10. Integer scoring fine; >1-decimal is false precision.\n- Impact: estimated NSM lift IF experiment wins. 10 = doubles NSM; 1 = barely measurable.\n- Confidence: P(experiment wins as hypothesized). 10 = replicates known win; 1 = wild guess.\n- Ease: inverse cost to ship+measure. 10 = afternoon, no eng; 1 = quarter-long.\n\n### Thresholds\n- ICE ≥ 8.0 → green-light immediately.\n- ICE 7.0-7.9 → next sprint. Default \"worth running\" cutoff.\n- ICE 5.0-6.9 → backlog. Re-score when context shifts.\n- ICE < 5.0 → reject or rework.\n\n### When/Then\n- 3+ experiments score ≥ 7 same week → parallel only if different funnel stages; else sequence (avoid interaction effects).\n- All cluster 5-7 → backlog stale; force brainstorming round on Impact-10 ideas.\n- Top-ICE all Ease=9 quick wins but NSM flat 4 weeks → optimizing locally. Force ≥1 high-Impact/low-Ease per sprint.\n- Confidence consistently > 8 → not exploring; bias toward more uncertain experiments.\n- Confidence consistently < 4 → guessing; invest in research before more experiments.\n\n### Anti-pattern\n- Chasing low-Ease wins that don't move NSM. Quick wins ≠ growth. Sean Ellis explicitly: ICE is for fast triage, NOT serious trade-off prioritization. Use RICE or WSJF for hard trade-offs.\n- Committee-average without dissent. Score independently, discuss outliers.\n- Anchoring Confidence on Impact. Score independently.\n- One person owning all three scores → consistent bias.\n\n### Variants\n- RICE = (Reach × Impact × Confidence) / Effort. Use when Reach varies widely.\n- WSJF = (User-Biz Value + Time Criticality + Risk Reduction) / Job Size. Cross-functional roadmap (SAFe).\n- Nexus stores all three. ICE = weekly growth triage; RICE = cross-segment audience; WSJF = roadmap vs eng capacity.\n\n### LLM scoring rules for ice_suggest\n- Impact from hypothesized NSM lift: +50% = 10, +25% = 8, +10% = 6, +5% = 4, +1% = 2, <1% = 1.\n- Confidence by similarity to historical wins in experiments table: direct replication = 9, similar pattern = 7, novel = 4, speculation = 2.\n- Ease from eng_effort_estimate_days: < 1d = 10, 1-3d = 8, 1w = 6, 2w = 4, > 1mo = 2.\n- Always show breakdown (I, C, E) + composite. Never return only the score.\n\n### Interpretation for chat\n- New experiment idea → score 1-10 per axis with one-line justification each.\n- Ranking a list → sort by ICE desc, tiebreak by Ease (cheaper first).\n- Refuse to score if Impact target metric unspecified. Ask \"lift on which metric?\" first.\n\n### Decision rules\n- ice_suggest auto-runs on any new hypothesis.\n- Top-3 surfaced in recommendations only if ICE ≥ 7 AND target metric matches an at-risk KPI from funnel_summary.`,\n },\n\n {\n name: 'AARRR Pirate Metrics',\n kind: 'framework',\n targets: ['funnel_summary', 'chat', 'correlations'],\n content: `Source: Dave McClure, 500 Startups, 2007. Five-stage user lifecycle for early-stage growth.\n\n### Stages\n1. Acquisition — user arrives. Metrics: sessions, signups, signup conv, CAC.\n2. Activation — first valuable experience (aha). Metrics: activation rate, time-to-activate, onboarding completion.\n3. Retention — user returns. Metrics: D1/D7/D30, WAU/MAU, churn.\n4. Referral — user invites. Metrics: viral coefficient K, invites/user, acceptance.\n5. Revenue — user pays. Metrics: ARPU, conv to paid, expansion, LTV.\n\n### Mapping to T-M-B funnel (Nexus)\n- TOFU = Acquisition + Activation\n- MOFU = Retention\n- BOFU = Referral + Revenue\nCompresses 5 stages into 3-tier display while preserving signal.\n\n### Stage health\n- Normalize each metric to [0, 100] vs industry benchmark:\n - higher better: score = clip(value / benchmark × 100, 0, 100)\n - lower better: score = clip(benchmark / value × 100, 0, 100)\n- Stage_health(s) = mean(normalized scores).\n\n### Thresholds\n- Stage health ≥ 75 = strong. 50-74 = OK. < 50 = bottleneck (flag).\n- AARRR weights (Nexus): A 0.20, A 0.20, R 0.25, R 0.15, R 0.20 (Retention emphasized).\n- T-M-B weights: TOFU 0.25, MOFU 0.40, BOFU 0.35.\n\n### When/Then\n- Single stage < 50 → THE bottleneck. Do NOT optimize others until fixed. Sequential.\n- Acquisition strong + Activation < 50 → leaky onboarding. Onboarding experiment.\n- Activation + Retention both < 50 → PMF issue, not marketing. Pause acquisition until fixed.\n- Retention strong + Revenue < 50 → monetization issue. Pricing/packaging review.\n- Referral < 50 + Revenue strong → leaving growth on table. Referral program experiment.\n- All 50-75 → balanced; pick biggest delta-to-benchmark for next experiment.\n\n### Correlation signals\n- Acquisition spike + Activation drop → traffic quality degraded (audience/creative mismatch).\n- Retention drop + Revenue stable → expansion masking churn.\n- Referral up + CAC down → viral loop firing; scale infra.\n- Activation up + Retention flat → fake activation event (definition too loose).\n\n### Anti-pattern\n- Optimizing acquisition before activation solid. Leaky bucket. McClure's original sin.\n- Picking activation event by ease-of-tracking, not aha-moment correlation with retention. Validate: users who hit X retain Z% vs Y% baseline.\n- AARRR as strict UX sequence — it's measurement framework, not journey map. Users can refer before activating.\n- Equal-weighting all stages. Retention compounds; weight higher.\n\n### Interpretation for chat\n- \"Where's the bottleneck?\" → stage health for all 5 (or T-M-B 3) with normalized scores. Identify single lowest, recommend that stage's top-ICE experiments.\n- Never propose Revenue optimization when Retention is the bottleneck. Sequence matters.\n- Use AARRR vocab in B2C/freemium. Use T-M-B/pipeline vocab in B2B sales-led.\n\n### Decision rules\n- funnel_summary returns stage_health(stage) + overall_health.\n- correlations engine scans stage-pair signals above and emits alerts.`,\n },\n\n {\n name: 'Funnel Health Scoring',\n kind: 'formula',\n targets: ['funnel_summary', 'recommendations', 'chat'],\n content: `Source: Nexus internal scoring (AARRR + GA4/HubSpot benchmarks). Used by api/intelligence.ts.\n\n### Formula\nFor each funnel stage s ∈ {TOFU, MOFU, BOFU}:\n- Stage_health(s) = mean(normalize(m_i, benchmark_i, direction_i)).\n- normalize(value, benchmark, direction):\n - 'up' (higher better): clip(value / benchmark × 100, 0, 100)\n - 'down' (lower better): clip(benchmark / value × 100, 0, 100)\n- Overall_health = 0.25 × TOFU + 0.40 × MOFU + 0.35 × BOFU.\n- Trend bonus: ±15 if delta_7d direction confirms benchmark improvement. Cap final at 100.\n\n### Metric → stage mapping (default)\n- TOFU: sessions, new_users, signup_rate, bounce_rate(↓), CPC(↓), impressions, CTR.\n- MOFU: activation_rate, D7_retention, MQL_count, lead_score_dist, time_to_activate(↓), trial_to_paid_rate.\n- BOFU: SQL_count, win_rate, avg_deal_size, MER, LTV/CAC, NRR, churn_rate(↓).\n\n### Benchmark anchors (B2B SaaS defaults)\nbounce_rate 50%, signup_conv 3%, activation_rate 35%, D7_retention 30%, trial_to_paid 15%, win_rate 25%, NRR 110%, monthly logo churn 2%, LTV/CAC 3.0.\n\n### Thresholds\n- Stage ≥ 75 = green. 50-74 = yellow. < 50 = red.\n- Overall ≥ 70 = healthy. 50-69 = mixed. < 50 = systemic issue.\n\n### When/Then\n- Stage < 50 → \"bottleneck\" alert; recommendations surfaces stage-specific experiments first.\n- Stage declines > 10 pts WoW → \"rapid decline\" alert even if absolute still > 50.\n- TOFU green + MOFU red → traffic quality OR onboarding. Audience refinement + activation experiment.\n- MOFU green + BOFU red → conversion/monetization. Pricing or sales-process experiment.\n- All green but overall flat → optimization plateau; new-channel exploration.\n- Trend bonus pushes 49→64 → label \"improving\" but use RAW score as action signal.\n\n### Trend bonus\n- delta_7d = today - 7d_ago. bonus = clip(delta_7d × 0.5, -15, +15). Reported = clip(raw + bonus, 0, 100).\n- Recommendations engine uses RAW for thresholds. Bonus is UI/digest only.\n\n### Decomposition\nWhen stage drops: delta_contribution(m_i) = (norm_today - norm_7d) × (1 / N_metrics). Surface top-2 negative contributors in digest.\n\n### Anti-pattern\n- Equal weighting across stages. Retention/BOFU compound; weight higher (0.40 MOFU, 0.35 BOFU).\n- Single-metric stage health (only bounce_rate for TOFU). ≥3 metrics per stage to reduce noise.\n- Single industry benchmark for multi-segment business. Use segment-specific when SMB/Ent mix > 30/70.\n- Confusing trend bonus with action signal. Bonus = narrative; raw = decisions.\n- Bench-marking against world-class for young companies — perpetual red flags.\n\n### Interpretation for chat\n- \"How's the funnel?\" → TOFU/MOFU/BOFU health (raw + reported), overall, top-2 declining metrics.\n- Always cite benchmark used. e.g. \"TOFU 62/100 because bounce_rate 58% vs benchmark 50%.\"\n- Overall > 70 but single stage < 50 → lead with bottleneck, not headline.\n\n### Decision rules\n- raw stage_health < 50 → focus for recommendations top-3.\n- delta_7d_health < -10 → push to standup regardless of absolute level.`,\n },\n\n {\n name: 'MQL/SQL Lead Scoring',\n kind: 'framework',\n targets: ['lead_scoring', 'chat', 'digest'],\n content: `Source: HubSpot / Marketo conventions, adapted for Nexus signals table.\n\n### MQL score formula\nSum activity points over rolling 30d window with decay:\n- +5 email open / passive page view\n- +15 link click (email or content)\n- +25 pricing / plans / cost mention or page visit\n- +20 gated content download (whitepaper, ebook, calculator)\n- +20 webinar registration or attendance\n- +10 repeat email open within 7d (max 3× = +30)\n- +10 return visit within 7d\n\n### SQL signals (boolean count)\n- Reply to outreach = 1; Demo request = 1; \"Talk to sales\" form = 1; Calendar event booked = 1; Direct inbound call = 1.\n\n### Qualification rules\n- mql_qualified = score ≥ 40 in 30-day window.\n- sql_qualified = sql_signal_count ≥ 2 OR any single signal in {demo_request, calendar_event} (single-signal qualifies — see intent_signal_weights).\n- Hot lead (outreach priority): score ≥ 75.\n- Cold (deprioritize): score < 15 AND no signals in 30d.\n\n### Decay\n- Signals > 14d old: × 0.5.\n- Signals > 30d old: × 0.25 (effectively drop).\n- Apply at query time; do not destroy raw signal records.\n\n### Account-level bonus (B2B)\n- +15 if ≥ 3 signals from same domain within 7d. Captures buying-committee activity.\n- +10 if second person from domain registers a signal within 14d.\n- Account score = max(individual_score) + bonus, surfaced separately.\n\n### When/Then\n- Score ≥ 40 + not contacted → MQL queue; SDR SLA 24h.\n- Score ≥ 75 (hot) → AE queue; outreach SLA 4h.\n- sql_qualified AND no meeting in 7d → escalate to RevOps; coverage gap.\n- Score plateaus 40-60 for > 21d → nurture sequence; not sales-ready.\n- Score spikes < 20 → > 60 in 7d → \"fast riser\" tag; priority outreach.\n- demo_request from low-score lead (< 20) → trust the demo; demo > behavioral score.\n\n### Anti-pattern\n- No decay → stale leads pollute MQL queue.\n- Single-metric scoring (only opens). Easily gamed.\n- MQL volume as KPI without MQL→SQL conversion. Volume without quality = wasted SDR time.\n- Manual overrides without logging. Kills model improvement loop.\n- Same rules for inbound vs outbound — outbound needs different weights.\n\n### Interpretation for chat\n- \"Is lead X hot?\" → score, signal breakdown by category, recency. Include time-since-last-signal.\n- MQL queue → sort by score desc, but surface \"fast risers\" first regardless of absolute score.\n- Never qualify a lead without signal data; if signals_loaded=false, ask to refresh integrations.\n\n### Decision rules (Nexus)\n- lead_scoring engine recomputes daily; top-25 by score surfaced.\n- digest \"Hot leads this week\" = score ≥ 75 OR sql_qualified=true.\n- standup \"New SQLs overnight\" = transitions to sql_qualified=true in last 24h.\n\n### Conversion benchmarks\n- MQL→SQL: 13% B2B SaaS (Forrester). < 8% = quality issue. > 25% = threshold too high.\n- SQL→Won: 20-30%. < 15% = qualification issue (false positives).`,\n },\n\n {\n name: 'Intent Signal Weights (B2B SaaS)',\n kind: 'benchmark',\n targets: ['lead_scoring', 'digest'],\n content: `Source: Nexus signal weights, calibrated from HubSpot/6sense intent conventions.\n\n### Weight table (signal | pts | tier)\n- email_open 5 passive | email_open_repeat (7d) +5 ea cap +15 passive | email_click 15 engaged | email_reply 25 high-intent\n- page_view (non-pricing) 3 passive | pricing_page_view 25 high-intent | pricing_keyword 30 high-intent\n- content_download 20 engaged | webinar_registration 20 engaged | webinar_attendance 25 high-intent\n- demo_request 50 SQL-tier | calendar_event (booked) 50 SQL-tier\n- linkedin_view (company) 10 engaged | linkedin_engagement 15 engaged | g2_review_view (category) 15 engaged\n- competitor_comparison_visit 30 high-intent | pricing_calculator_use 35 high-intent | contract_template_download 35 high-intent | support_ticket_pre_sale 20 engaged\n\n### Tier rules\n- passive: alone insufficient; ≥ 3 needed.\n- engaged: contributes meaningfully; doesn't qualify alone.\n- high-intent: 2+ in 14d = MQL+; surface to SDR.\n- SQL-tier: SINGLE demo_request OR calendar_event = SQL.\n\n### Aggregation\n- Per lead: sum points over rolling 30d with decay.\n- Per account: max(lead_score) + account_bonus.\n- Buckets: 0-7d full, 7-14d ×0.5, 14-30d ×0.25, > 30d drop.\n\n### When/Then\n- demo_request OR calendar_event fires → auto-create SQL; sales SLA 4h.\n- pricing_page_view + competitor_comparison_visit within 7d same lead → in-market; assign AE now.\n- pricing_calculator_use → very strong intent; surface in standup \"high-intent activity overnight.\"\n- email_open_repeat hits cap (+15) but no clicks → interested in subject lines, not content. Refresh creative.\n- linkedin_view spikes from account with no other signals → research stage; tag, don't outreach yet.\n\n### Industry calibration\n- Enterprise (ACV > $50k): calendar_event 50→35; contract_template_download 35→50.\n- SMB / PLG (ACV < $5k): pricing_calculator_use 35→50; demo_request 50→30.\n- Do NOT use single weight table across heterogeneous segments.\n\n### Combination signals\n- pricing_page_view + competitor_comparison_visit within 7d → +25 (research depth).\n- demo_request + content_download within 14d prior → \"primed buyer\"; expedite.\n- email_reply + linkedin_engagement same person → multi-channel; high-quality lead.\n\n### Anti-pattern\n- Counting page_view as engaged. Drive-by traffic; keep passive (3 pts).\n- Equal-weighting demo_request and email_open. Off by 10x in actual intent.\n- Ignoring company-level signal (leads as independent). B2B is buying-committee.\n- Static weights for years. Recalibrate quarterly against actual SQL→Close conversion.\n\n### Interpretation for chat\n- List lead's signals in TIER order (SQL > high-intent > engaged > passive), not chronological.\n- \"Should I reach out?\" → any SQL-tier in 7d = yes now. High-intent stack ≥ 2 in 14d = yes today. Else nurture.\n- Always cite specific signals driving the score; opacity destroys trust.`,\n },\n\n {\n name: 'Engagement Velocity',\n kind: 'formula',\n targets: ['lead_scoring', 'digest', 'standup'],\n content: `Source: SaaS sales-cycle analytics (Bridge Group, Salesforce velocity reports).\n\n### Formula\n- velocity = mean(SQL_qualified_at - MQL_qualified_at) across leads transitioning in a period.\n- Use MEDIAN for skew-resistance; report both mean and median.\n- Compute per segment (SMB / mid / enterprise) — blended hides segment differences.\n\n### Variants\n- mql_to_sql_velocity = mean(SQL_at - MQL_at) — primary metric.\n- sql_to_close_velocity = mean(closed_won_at - SQL_at).\n- first_touch_to_close = mean(closed_won_at - first_signal_at) — full cycle.\n- mql_to_close = mql_to_sql + sql_to_close.\n\n### Benchmarks (B2B SaaS, ACV $5k-$50k)\n- mql_to_sql_velocity median: 14-21 days. < 14d = fast. > 30d = slow.\n- sql_to_close_velocity median: 30-60 days. Varies sharply with ACV.\n- Total cycle: 60-120 days SMB; 6-12 months enterprise.\n- PLG / freemium: mql_to_sql can be < 7 days if product usage drives qualification.\n\n### Thresholds\n- velocity < 7d → review qualification rigor; likely false positives.\n- velocity 7-14d → fast + healthy for SMB/mid-market.\n- velocity 14-30d → standard; benchmark range.\n- velocity > 30d → AE engagement quality or fit issue. Investigate.\n- velocity > 60d → process broken or wrong ICP being scored.\n\n### When/Then\n- velocity > 30d AND MQL→SQL < 13% → AE engagement quality issue. SDR coaching, faster touch SLA.\n- velocity > 30d AND MQL→SQL > 20% → leads qualify slowly. ICP fit good, motion slow. Shorter discovery, multi-thread sooner.\n- velocity < 7d AND MQL→SQL > 40% → qualification too loose. Raise sql_signals 2→3; audit recent SQLs.\n- velocity declines WoW → process improving OR threshold loosening. Pair with conversion to disambiguate.\n- velocity spikes up → likely SDR capacity / vacation. Check headcount before assuming process failure.\n\n### Decomposition\nvelocity = Σ stage_dwell: mql_to_first_touch + first_touch_to_discovery + discovery_to_sql. Surface LONGEST sub-stage as bottleneck.\n\n### Cohort version\nPer MQL_at week-cohort. Trailing 8-week table shows trend. Recent 2 cohorts > 30d while prior were 18d → regression; investigate last 2 weeks of process changes.\n\n### Anti-pattern\n- Averaging across segments. SMB 10d + Enterprise 90d blended = 50d, useless.\n- Counting only closed leads → excludes slow/stalled, biases low. Include all SQL-qualified, censor at 90d.\n- Minimize-as-target without watching conversion. Faster ≠ better if quality drops.\n- Computing weekly with n < 20. Use 4-week rolling window.\n\n### Interpretation for chat\n- \"Are leads moving fast enough?\" → mql_to_sql median vs benchmark, with conversion context. Two numbers, not one.\n- velocity changes > 20% WoW → standup highlight with cohort decomposition.\n- Refuse single velocity number if segment mix unknown; ask for segment.\n\n### Pipeline velocity (revenue formulation)\npipeline_velocity = (num_opps × win_rate × avg_deal_size) / sales_cycle_days. Dollars-per-day through pipeline. Use for revenue forecasting; engagement velocity (above) is for lead-flow diagnosis.\n\n### Decision rules (Nexus)\n- velocity > 30d for 2 consecutive 4-week windows → recommendations: \"AE engagement quality review.\"\n- velocity < 7d AND mql→sql > 40% → digest \"Qualification rigor: false-positive risk.\"\n- standup \"Yesterday's SQLs\" includes individual transition times; outliers (> 2× median) flagged.`,\n },\n];\n",[],[365,778],"5953099c-5b19-11f1-98e7-ee650bb23af1","llm.ts","31190010-56ca-11f1-8403-ee650bb23af1",468,["D",1780029985598],34,"import { sqlite } from \"../db.ts\";\nimport { isSafePublicUrl } from \"./integrations/_shared.ts\";\nimport type { LLMConfig, ChatMessage } from \"../types.ts\";\n\n// --- Defaults -----------------------------------------------------------------\n\nconst DEFAULT_CONFIG: LLMConfig = {\n id: \"default\",\n provider: \"anthropic\",\n model: \"claude-sonnet-4-6\",\n api_key: \"\",\n endpoint: \"\",\n};\n\n// Provider - default model. Used to auto-correct on provider switch so users don't\n// accidentally send \"claude-sonnet-4-6\" to Gemini (which 400s).\nconst PROVIDER_DEFAULT_MODEL: Record<string, string> = {\n anthropic: \"claude-sonnet-4-6\",\n openai: \"gpt-4o\",\n gemini: \"gemini-3-pro-preview\",\n custom: \"your-model\",\n};\nconst VALID_PROVIDERS = new Set([\"anthropic\", \"openai\", \"gemini\", \"custom\"]);\n\n// Map a provider name to its dedicated key column on llm_config.\nconst providerKeyCol = (p: string): string => {\n switch (p) {\n case \"anthropic\": return \"api_key_anthropic\";\n case \"openai\": return \"api_key_openai\";\n case \"gemini\": return \"api_key_gemini\";\n case \"custom\": return \"api_key_custom\";\n default: return \"api_key_anthropic\";\n }\n};\n\n// --- Helpers ------------------------------------------------------------------\n\nfunction rowToConfig(columns: string[], row: any[]): LLMConfig & { has_key?: Record<string, boolean> } {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n const provider = obj.provider ?? DEFAULT_CONFIG.provider;\n // Resolve api_key: prefer the per-provider column, fall back to legacy single api_key.\n const perProviderKey = obj[providerKeyCol(provider)];\n const api_key = (perProviderKey && String(perProviderKey)) || obj.api_key || \"\";\n const endpoint = provider === \"custom\" ? (obj.endpoint_custom || obj.endpoint || \"\") : (obj.endpoint || \"\");\n return {\n id: obj.id ?? \"default\",\n provider,\n model: obj.model ?? DEFAULT_CONFIG.model,\n api_key,\n endpoint,\n has_key: {\n anthropic: !!(obj.api_key_anthropic || (provider === \"anthropic\" && obj.api_key)),\n openai: !!(obj.api_key_openai || (provider === \"openai\" && obj.api_key)),\n gemini: !!(obj.api_key_gemini || (provider === \"gemini\" && obj.api_key)),\n custom: !!(obj.api_key_custom || (provider === \"custom\" && obj.api_key)),\n },\n } as any;\n}\n\n// --- DB operations ------------------------------------------------------------\n\nexport async function getLLMConfig(): Promise<LLMConfig> {\n const result = await sqlite.execute(\n `SELECT id, provider, model, api_key, endpoint,\n api_key_anthropic, api_key_openai, api_key_gemini, api_key_custom, endpoint_custom\n FROM llm_config WHERE id = 'default'`\n );\n\n if (!result.rows || result.rows.length === 0) {\n return { ...DEFAULT_CONFIG };\n }\n\n return rowToConfig(result.columns, result.rows[0]);\n}\n\nexport async function saveLLMConfig(\n config: Partial<LLMConfig>\n): Promise<LLMConfig> {\n const existing = await getLLMConfig();\n\n // Validate provider; fall back to existing if caller sent garbage.\n let provider = (config.provider ?? existing.provider) as string;\n if (!VALID_PROVIDERS.has(provider)) provider = existing.provider;\n\n // Auto-correct model when provider changes and caller didn't pick a sensible one.\n // Prevents \"model not found\" 400s when switching anthropic - gemini.\n let model = config.model ?? existing.model;\n if (provider !== existing.provider) {\n const userPickedNewModel = config.model && config.model !== existing.model;\n if (!userPickedNewModel) {\n model = PROVIDER_DEFAULT_MODEL[provider] ?? existing.model;\n }\n }\n\n // Per-provider key: if caller supplied api_key, route it to that provider's column.\n // Other providers' keys are preserved across switches.\n const incomingKey = config.api_key;\n const keyCol = providerKeyCol(provider);\n const endpoint = provider === \"custom\" ? (config.endpoint ?? existing.endpoint ?? \"\") : (existing.endpoint ?? \"\");\n\n const now = new Date().toISOString();\n // Upsert: keep existing per-provider columns, only overwrite the one(s) the user changed.\n await sqlite.execute(\n `INSERT INTO llm_config (id, provider, model, api_key, endpoint, ${keyCol}, endpoint_custom, updated_at)\n VALUES ('default', ?, ?, ?, ?, ?, ?, ?)\n ON CONFLICT(id) DO UPDATE SET\n provider = excluded.provider,\n model = excluded.model,\n api_key = excluded.api_key,\n endpoint = excluded.endpoint,\n ${keyCol} = COALESCE(NULLIF(excluded.${keyCol}, ''), llm_config.${keyCol}),\n endpoint_custom = COALESCE(NULLIF(excluded.endpoint_custom, ''), llm_config.endpoint_custom),\n updated_at = excluded.updated_at`,\n [\n provider,\n model,\n incomingKey ?? existing.api_key ?? \"\",\n endpoint,\n incomingKey ?? \"\",\n provider === \"custom\" ? endpoint : \"\",\n now,\n ]\n );\n\n return await getLLMConfig();\n}\n\n// --- LLM call -----------------------------------------------------------------\n\nexport async function callLLM(\n config: LLMConfig,\n messages: ChatMessage[],\n options?: {\n systemPrompt?: string;\n tools?: any[];\n maxTokens?: number;\n temperature?: number; // 0..1 - passed to provider\n frequencyPenalty?: number; // 0..2 - OpenAI/custom only (Anthropic ignores)\n }\n): Promise<{ content: string; tool_calls?: any[] }> {\n const maxTokens = options?.maxTokens ?? 4096;\n const systemPrompt = options?.systemPrompt;\n const tools = options?.tools;\n const temperature = options?.temperature;\n const frequencyPenalty = options?.frequencyPenalty;\n\n if (config.provider === \"anthropic\") {\n return callAnthropic(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"openai\") {\n return callOpenAI(config, \"https://api.openai.com/v1\", messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else if (config.provider === \"gemini\") {\n return callGemini(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"custom\") {\n const endpoint = config.endpoint?.replace(/\\/+$/, \"\") ?? \"\";\n // SSRF guard: a custom LLM endpoint is user-supplied and fetched server-side with the API key.\n if (!isSafePublicUrl(endpoint)) {\n throw new Error(\"Custom LLM endpoint must be a public https:// URL\");\n }\n return callOpenAI(config, endpoint, messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else {\n throw new Error(`Unknown LLM provider: ${config.provider}`);\n }\n}\n\n// --- Streaming LLM call -------------------------------------------------------\n\nexport async function* callLLMStream(\n config: LLMConfig,\n messages: ChatMessage[],\n options?: {\n systemPrompt?: string;\n tools?: any[];\n maxTokens?: number;\n temperature?: number;\n frequencyPenalty?: number;\n }\n): AsyncGenerator<string> {\n const maxTokens = options?.maxTokens ?? 4096;\n const systemPrompt = options?.systemPrompt;\n const tools = options?.tools;\n const temperature = options?.temperature;\n const frequencyPenalty = options?.frequencyPenalty;\n\n if (config.provider === \"anthropic\") {\n yield* callAnthropicStream(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"openai\") {\n yield* callOpenAIStream(config, \"https://api.openai.com/v1\", messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else if (config.provider === \"gemini\") {\n yield* callGeminiStream(config, messages, systemPrompt, tools, maxTokens, temperature);\n } else if (config.provider === \"custom\") {\n const endpoint = config.endpoint?.replace(/\\/+$/, \"\") ?? \"\";\n if (!isSafePublicUrl(endpoint)) {\n throw new Error(\"Custom LLM endpoint must be a public https:// URL\");\n }\n yield* callOpenAIStream(config, endpoint, messages, systemPrompt, tools, maxTokens, temperature, frequencyPenalty);\n } else {\n throw new Error(`Unknown LLM provider: ${config.provider}`);\n }\n}\n\nasync function* parseSseStream(body: ReadableStream<Uint8Array>, onChunk: (parsed: any) => string | null): AsyncGenerator<string> {\n const reader = body.getReader();\n const dec = new TextDecoder();\n let buf = \"\";\n const chunks: string[] = [];\n\n const flush = async function* () {\n while (chunks.length) {\n yield chunks.shift()!;\n }\n };\n\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) break;\n buf += dec.decode(value, { stream: true });\n const lines = buf.split(\"\\n\");\n buf = lines.pop()!;\n for (const line of lines) {\n if (!line.startsWith(\"data: \")) continue;\n const d = line.slice(6).trim();\n if (d === \"[DONE]\") return;\n try {\n const j = JSON.parse(d);\n const text = onChunk(j);\n if (text) yield text;\n } catch { /* skip malformed SSE lines */ }\n }\n }\n } finally {\n reader.cancel().catch(() => {});\n }\n}\n\nasync function* callAnthropicStream(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): AsyncGenerator<string> {\n const body: Record<string, any> = {\n model: config.model ?? \"claude-sonnet-4-6\",\n max_tokens: maxTokens,\n messages: toAnthropicMessages(messages),\n stream: true,\n };\n if (typeof temperature === \"number\") body.temperature = Math.max(0, Math.min(1, temperature));\n if (systemPrompt) body.system = systemPrompt;\n if (tools && tools.length > 0) {\n body.tools = tools.map((t) => {\n if (t.type === \"function\") {\n return { name: t.function.name, description: t.function.description ?? \"\", input_schema: t.function.parameters ?? { type: \"object\", properties: {} } };\n }\n return t;\n });\n }\n\n const res = await fetch(\"https://api.anthropic.com/v1/messages\", {\n method: \"POST\",\n headers: { \"x-api-key\": config.api_key, \"anthropic-version\": \"2023-06-01\", \"Content-Type\": \"application/json\" },\n body: JSON.stringify(body),\n });\n if (!res.ok) throw new Error(`Anthropic ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n if (j.type === \"content_block_delta\" && j.delta?.text) return j.delta.text;\n return null;\n });\n}\n\nasync function* callOpenAIStream(\n config: LLMConfig,\n baseUrl: string,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number,\n frequencyPenalty?: number\n): AsyncGenerator<string> {\n const openAIMessages: any[] = [];\n if (systemPrompt) openAIMessages.push({ role: \"system\", content: systemPrompt });\n for (const m of messages as any[]) {\n if (m.role === \"assistant\" && m.tool_calls) {\n const msg: any = { role: \"assistant\", content: m.content ?? null, tool_calls: m.tool_calls };\n if (m.reasoning_content) msg.reasoning_content = m.reasoning_content;\n openAIMessages.push(msg);\n } else if (m.role === \"tool\") {\n openAIMessages.push({ role: \"tool\", tool_call_id: m.tool_call_id, content: String(m.content ?? \"\") });\n } else {\n openAIMessages.push({ role: m.role, content: m.content ?? \"\" });\n }\n }\n\n const body: Record<string, any> = {\n model: config.model ?? \"gpt-4o\",\n messages: openAIMessages,\n max_tokens: maxTokens,\n temperature: typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 0.7,\n stream: true,\n };\n if (typeof frequencyPenalty === \"number\") body.frequency_penalty = Math.max(0, Math.min(2, frequencyPenalty));\n if (tools && tools.length > 0) { body.tools = tools; body.tool_choice = \"auto\"; }\n\n const res = await fetch(`${baseUrl}/chat/completions`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${config.api_key}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify(body),\n });\n if (!res.ok) throw new Error(`OpenAI ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n const text = j.choices?.[0]?.delta?.content;\n if (text) return text;\n return null;\n });\n}\n\nasync function* callGeminiStream(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): AsyncGenerator<string> {\n const model = config.model ?? \"gemini-3-pro-preview\";\n // Use ?alt=sse for server-sent events streaming\n const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:streamGenerateContent?alt=sse`;\n\n const tcNameMap: Record<string, string> = {};\n for (const m of messages as any[]) {\n if (m.tool_calls) for (const tc of m.tool_calls) tcNameMap[tc.id] = tc.function.name;\n }\n\n const contents: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"system\") continue;\n if (m.role === \"tool\") {\n const fnName = tcNameMap[m.tool_call_id] ?? m.tool_call_id ?? \"tool\";\n const part = { functionResponse: { id: (m as any).tool_call_id, name: fnName, response: { content: m.content } } };\n const last = contents[contents.length - 1];\n if (last?.role === \"user\") last.parts.push(part);\n else contents.push({ role: \"user\", parts: [part] });\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const parts: any[] = [];\n if (m.content) parts.push({ text: m.content });\n for (const tc of m.tool_calls) {\n let args: any = {};\n try { args = typeof tc.function.arguments === \"string\" ? JSON.parse(tc.function.arguments) : (tc.function.arguments ?? {}); } catch { args = {}; }\n parts.push({ functionCall: { id: tc.id, name: tc.function.name, args } });\n }\n contents.push({ role: \"model\", parts });\n } else {\n const role = m.role === \"assistant\" ? \"model\" : \"user\";\n contents.push({ role, parts: [{ text: m.content }] });\n }\n }\n\n const genConfig: Record<string, any> = { maxOutputTokens: maxTokens };\n genConfig.temperature = typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 1.0;\n genConfig.thinkingConfig = { thinkingLevel: \"medium\" };\n\n const reqBody: Record<string, any> = { contents, generationConfig: genConfig };\n if (systemPrompt) reqBody.systemInstruction = { parts: [{ text: systemPrompt }] };\n if (tools && tools.length > 0) {\n reqBody.tools = [{ functionDeclarations: tools.map(t => { const fn = t.type === \"function\" ? t.function : t; return { name: fn.name, description: fn.description ?? \"\", parameters: stripUnsupported(fn.parameters ?? { type: \"object\", properties: {} }) }; }) }];\n reqBody.toolConfig = { functionCallingConfig: { mode: \"AUTO\" } };\n }\n\n const res = await fetch(url, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\", \"x-goog-api-key\": config.api_key },\n body: JSON.stringify(reqBody),\n });\n if (!res.ok) throw new Error(`Gemini ${res.status}: ${await res.text()}`);\n\n yield* parseSseStream(res.body!, (j) => {\n const text = j.candidates?.[0]?.content?.parts?.[0]?.text;\n if (text) return text;\n return null;\n });\n}\n\n// --- Anthropic ----------------------------------------------------------------\n\nfunction toAnthropicMessages(messages: ChatMessage[]): any[] {\n const result: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"tool\") {\n const block = { type: \"tool_result\", tool_use_id: m.tool_call_id, content: m.content };\n const prev = result[result.length - 1];\n if (prev && prev.role === \"user\" && Array.isArray(prev.content)) {\n prev.content.push(block);\n } else {\n result.push({ role: \"user\", content: [block] });\n }\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const content: any[] = [];\n if (m.content) content.push({ type: \"text\", text: m.content });\n for (const tc of m.tool_calls) {\n let input: any = {};\n try {\n input = typeof tc.function.arguments === \"string\"\n ? JSON.parse(tc.function.arguments) : tc.function.arguments ?? {};\n } catch { input = {}; }\n content.push({ type: \"tool_use\", id: tc.id, name: tc.function.name, input });\n }\n result.push({ role: \"assistant\", content });\n } else {\n result.push({ role: m.role, content: m.content });\n }\n }\n return result;\n}\n\nasync function callAnthropic(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n const body: Record<string, any> = {\n model: config.model ?? \"claude-sonnet-4-6\",\n max_tokens: maxTokens,\n messages: toAnthropicMessages(messages),\n };\n if (typeof temperature === \"number\") body.temperature = Math.max(0, Math.min(1, temperature));\n\n if (systemPrompt) {\n body.system = systemPrompt;\n }\n\n if (tools && tools.length > 0) {\n // Convert OpenAI-style tools to Anthropic format if needed\n body.tools = tools.map((t) => {\n if (t.type === \"function\") {\n return {\n name: t.function.name,\n description: t.function.description ?? \"\",\n input_schema: t.function.parameters ?? { type: \"object\", properties: {} },\n };\n }\n // Already in Anthropic format\n return t;\n });\n }\n\n // 25s timeout per LLM call (see callOpenAI for rationale).\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(\"https://api.anthropic.com/v1/messages\", {\n method: \"POST\",\n headers: {\n \"x-api-key\": config.api_key,\n \"anthropic-version\": \"2023-06-01\",\n \"Content-Type\": \"application/json\",\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`Anthropic API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n\n // Extract text content and tool use blocks\n let textContent = \"\";\n const toolCalls: any[] = [];\n\n if (Array.isArray(data.content)) {\n for (const block of data.content) {\n if (block.type === \"text\") {\n textContent += block.text;\n } else if (block.type === \"tool_use\") {\n toolCalls.push({\n id: block.id,\n type: \"function\",\n function: {\n name: block.name,\n arguments: JSON.stringify(block.input),\n },\n });\n }\n }\n }\n\n return {\n content: textContent,\n tool_calls: toolCalls.length > 0 ? toolCalls : undefined,\n };\n}\n\n// --- OpenAI / OpenAI-compatible ----------------------------------------------\n\nasync function callOpenAI(\n config: LLMConfig,\n baseUrl: string,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number,\n frequencyPenalty?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n const openAIMessages: any[] = [];\n\n if (systemPrompt) {\n openAIMessages.push({ role: \"system\", content: systemPrompt });\n }\n\n for (const m of messages as any[]) {\n if (m.role === \"assistant\" && m.tool_calls) {\n const msg: any = { role: \"assistant\", content: m.content ?? null, tool_calls: m.tool_calls };\n // DeepSeek thinking mode: reasoning_content must be echoed back for multi-turn\n if (m.reasoning_content) msg.reasoning_content = m.reasoning_content;\n openAIMessages.push(msg);\n } else if (m.role === \"tool\") {\n openAIMessages.push({ role: \"tool\", tool_call_id: m.tool_call_id, content: String(m.content ?? \"\") });\n } else {\n openAIMessages.push({ role: m.role, content: m.content ?? \"\" });\n }\n }\n\n const body: Record<string, any> = {\n model: config.model ?? \"gpt-4o\",\n messages: openAIMessages,\n max_tokens: maxTokens,\n temperature: typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 0.7,\n };\n if (typeof frequencyPenalty === \"number\") body.frequency_penalty = Math.max(0, Math.min(2, frequencyPenalty));\n\n if (tools && tools.length > 0) {\n body.tools = tools;\n body.tool_choice = \"auto\";\n }\n\n // Hard 25s timeout per LLM call. Without this, a slow/unreachable provider hangs\n // the whole chat handler past Val Town's 60s execution cap and triggers a pause.\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(`${baseUrl}/chat/completions`, {\n method: \"POST\",\n headers: {\n Authorization: `Bearer ${config.api_key}`,\n \"Content-Type\": \"application/json\",\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`OpenAI API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n const choice = data.choices?.[0];\n const message = choice?.message;\n\n const content: string = message?.content ?? \"\";\n const tool_calls: any[] | undefined =\n message?.tool_calls && message.tool_calls.length > 0\n ? message.tool_calls\n : undefined;\n // DeepSeek thinking mode: preserve reasoning_content so it can be echoed back in multi-turn\n const reasoning_content: string | undefined = message?.reasoning_content ?? undefined;\n\n return { content, tool_calls, reasoning_content };\n}\n\n// --- Gemini (native REST API) -------------------------------------------------\n\n// Gemini rejects additionalProperties and other JSON Schema fields it doesn't know\nfunction stripUnsupported(schema: any): any {\n if (!schema || typeof schema !== \"object\") return schema;\n const { additionalProperties, $schema, ...rest } = schema;\n if (rest.properties) {\n rest.properties = Object.fromEntries(\n Object.entries(rest.properties).map(([k, v]) => [k, stripUnsupported(v)])\n );\n }\n if (rest.items) rest.items = stripUnsupported(rest.items);\n return rest;\n}\n\nasync function callGemini(\n config: LLMConfig,\n messages: ChatMessage[],\n systemPrompt?: string,\n tools?: any[],\n maxTokens = 4096,\n temperature?: number\n): Promise<{ content: string; tool_calls?: any[] }> {\n // Gemini 3 migration: default model is now gemini-3-pro-preview (gemini-1.5-flash deprecated).\n // Auth moved from ?key= query to the x-goog-api-key header (canonical).\n const model = config.model ?? \"gemini-3-pro-preview\";\n const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:generateContent`;\n\n // Build tool_call_id -> function name map so functionResponse can use the name\n const tcNameMap: Record<string, string> = {};\n for (const m of messages as any[]) {\n if (m.tool_calls) {\n for (const tc of m.tool_calls) {\n tcNameMap[tc.id] = tc.function.name;\n }\n }\n }\n\n const contents: any[] = [];\n for (const m of messages as any[]) {\n if (m.role === \"system\") continue;\n\n if (m.role === \"tool\") {\n const fnName = tcNameMap[m.tool_call_id] ?? m.tool_call_id ?? \"tool\";\n // id must match the corresponding FunctionCall id (required in Gemini 3.x)\n const part = { functionResponse: { id: (m as any).tool_call_id, name: fnName, response: { content: m.content } } };\n const last = contents[contents.length - 1];\n if (last?.role === \"user\") {\n last.parts.push(part);\n } else {\n contents.push({ role: \"user\", parts: [part] });\n }\n } else if (m.role === \"assistant\" && m.tool_calls?.length) {\n const parts: any[] = [];\n if (m.content) parts.push({ text: m.content });\n for (const tc of m.tool_calls) {\n let args: any = {};\n try { args = typeof tc.function.arguments === \"string\" ? JSON.parse(tc.function.arguments) : (tc.function.arguments ?? {}); } catch { args = {}; }\n const fcPart: any = { functionCall: { id: tc.id, name: tc.function.name, args } };\n // Echo thoughtSignature exactly as received - required by Gemini 2.x thinking models\n if (tc._gemini_thought_signature) fcPart.thoughtSignature = tc._gemini_thought_signature;\n parts.push(fcPart);\n }\n contents.push({ role: \"model\", parts });\n } else {\n const role = m.role === \"assistant\" ? \"model\" : \"user\";\n contents.push({ role, parts: [{ text: m.content }] });\n }\n }\n\n // Gemini 3 defaults: temperature 1.0 (research confirms lowering it degrades reasoning).\n // We honor an explicit caller override only when in [0,2] (Gemini 3 expanded range).\n const genConfig: Record<string, any> = { maxOutputTokens: maxTokens };\n genConfig.temperature = typeof temperature === \"number\" ? Math.max(0, Math.min(2, temperature)) : 1.0;\n // Gemini 3 introduced thinkingLevel (\"minimal\"|\"low\"|\"medium\"|\"high\") replacing the\n // numeric thinking_budget. We default to \"medium\" - Gemini 3's reasoning mode.\n genConfig.thinkingConfig = { thinkingLevel: \"medium\" };\n\n const body: Record<string, any> = {\n contents,\n generationConfig: genConfig,\n };\n\n if (systemPrompt) {\n body.systemInstruction = { parts: [{ text: systemPrompt }] };\n }\n\n if (tools && tools.length > 0) {\n body.tools = [{\n functionDeclarations: tools.map(t => {\n const fn = t.type === \"function\" ? t.function : t;\n return { name: fn.name, description: fn.description ?? \"\", parameters: stripUnsupported(fn.parameters ?? { type: \"object\", properties: {} }) };\n }),\n }];\n body.toolConfig = { functionCallingConfig: { mode: \"AUTO\" } };\n }\n\n // 25s timeout per LLM call (see callOpenAI for rationale).\n const ac = new AbortController();\n const to = setTimeout(() => ac.abort(), 25_000);\n let res: Response;\n try {\n res = await fetch(url, {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n // Gemini 3 canonical auth - header replaces the legacy ?key= query param.\n \"x-goog-api-key\": config.api_key,\n },\n body: JSON.stringify(body),\n signal: ac.signal,\n });\n } finally {\n clearTimeout(to);\n }\n\n if (!res.ok) {\n const errText = await res.text();\n throw new Error(`Gemini API error ${res.status}: ${errText}`);\n }\n\n const data = await res.json();\n const parts = data.candidates?.[0]?.content?.parts ?? [];\n\n let textContent = \"\";\n const toolCalls: any[] = [];\n\n for (const part of parts) {\n if (part.text) textContent += part.text;\n else if (part.functionCall) {\n // Use the model's own id - must be echoed exactly in the functionResponse\n const tc: any = {\n id: part.functionCall.id ?? `gemini_${Date.now()}_${toolCalls.length}`,\n type: \"function\",\n function: {\n name: part.functionCall.name,\n arguments: JSON.stringify(part.functionCall.args ?? {}),\n },\n };\n // Preserve thoughtSignature (Part-level field) - required by Gemini 2.x thinking models\n const sig = part.thoughtSignature ?? part.thought_signature;\n if (sig) tc._gemini_thought_signature = sig;\n toolCalls.push(tc);\n }\n }\n\n return { content: textContent, tool_calls: toolCalls.length > 0 ? toolCalls : undefined };\n}\n",[],[365,786],"d4e758a2-5d61-11f1-be46-1607ee4eb77e","knowledge_seed.ts","d4e75974-5d61-11f1-be46-1607ee4eb77e",579,["D",1780281019016],"// W2 — Knowledge seed runner (Part 4 + Part 5 corrections).\n// `math_knowledge_seed.ts` only DEFINES MATH_KNOWLEDGE_SEEDS — nothing ever inserted it,\n// and the growth-model pack (Part 4) had no home. This module is the missing wire: it\n// inserts both packs into knowledge_files on boot, idempotently, gated by a version key in\n// app_config so it runs once (and re-runs only when the pack version bumps).\nimport { sqlite } from \"../db.ts\";\nimport { MATH_KNOWLEDGE_SEEDS } from \"./math_knowledge_seed.ts\";\n\nconst SEED_VERSION = 2; // bump to force a re-seed of new/changed rows\n\ninterface SeedRow { name: string; kind: string; targets: string[]; content: string; source: string }\n\n// Part 4 — growth-model mental models. Dense rules + thresholds + decision heuristics, ending\n// in an \"App wiring\" line so the AI knows how the engine should use each.\nconst GROWTH_MODEL_SEEDS: Array<{ name: string; targets: string[]; content: string }> = [\n {\n name: \"Growth loops > funnels\",\n targets: [\"chat\", \"recommendations\", \"correlations\"],\n content: `### Growth loops beat funnels\n- Funnels grow LINEARLY (must be re-fed forever); loops COMPOUND (output re-feeds input). Always ask \"how does one cohort produce the next?\"\n- Common SaaS loops: Content (publish → rank → traffic → signups → UGC/links → rank), Referral/Viral (users invite users; k-factor = invites × conversion-rate), Paid (spend → customers → revenue → reinvest; compounds only if payback < cycle), Sales (win → case study/referral → win).\n- A loop is healthy when its coefficient ≥ 1 over a reasonable cycle; below 1 it decays without external feeding.\n- App wiring: when recommending, prefer actions that strengthen a loop's weakest step over one-off funnel bumps; annotate correlations with which loop they belong to.`,\n },\n {\n name: \"Retention is the foundation\",\n targets: [\"funnel_summary\", \"kpis\", \"recommendations\", \"cohort_ltv\", \"retention\"],\n content: `### Retention is the foundation (the silent killer)\n- Retention compounds and is slow to reveal (judge over ≥12 months). Poor retention silently caps every acquisition dollar.\n- Watch the retention CURVE SHAPE: it must FLATTEN (a stable plateau = product-market fit); a curve trending to zero = leaky bucket → fix before scaling acquisition.\n- Benchmarks: GRR ≥ 90% (good); NRR ≥ 110–120% world-class at scale; < 100% NRR at scale = structural stickiness problem.\n- Fix order: activation → engagement (habit/loop) → resurrection → monetization expansion, before pouring in top-of-funnel.\n- App wiring: make NRR/GRR a North Star input; if the retention curve isn't flattening, down-rank acquisition recommendations.`,\n },\n {\n name: \"North Star + 3–5 inputs\",\n targets: [\"kpis\", \"recommendations\", \"chat\"],\n content: `### North Star Metric + 3–5 inputs\n- NSM = a LEADING indicator of value delivered (not lagging revenue). It must be movable via 3–5 input metrics that teams influence daily and that ladder up.\n- Canonical input shape: Breadth (# active) × Depth (actions/user) × Frequency (sessions/period) × Efficiency (conversion / time-to-value).\n- Avoid vanity NSMs (registered users, pageviews) and pure-lagging NSMs (MRR).\n- App wiring: require the NSM to declare inputs; show each input's contribution + which is the current constraint; pace on_track by input trajectory, not a straight line.`,\n },\n {\n name: \"AARRR as a diagnostic lens\",\n targets: [\"funnel_summary\", \"correlations\"],\n content: `### AARRR as a diagnostic lens (paired with loops)\n- Use Acquisition → Activation → Retention → Referral → Revenue to LOCATE the constraint; then attach a LOOP to relieve it.\n- Diagnose in reverse: weak Revenue often traces to weak Activation/Retention upstream — fix upstream first.\n- App wiring: keep the AARRR tag on every metric, but report the BINDING CONSTRAINT stage, not an average \"health\".`,\n },\n {\n name: \"Signal-based / account-based GTM (MQL is dead)\",\n targets: [\"lead_scoring\", \"chat\", \"recommendations\", \"accounts\"],\n content: `### Signal-based / account-based GTM\n- MQL is a LAGGING indicator; score ACCOUNTS, not individuals (B2B buys in committees).\n- Lead/Account score = FIT × INTENT. Fit = ICP firmographics (industry, size, role). Intent = TIME-DECAYED weighted signals (pricing visit, demo/proposal ask, reply, repeat visits, multiple contacts from one domain).\n- Apply DECAY (intent rots in days/weeks) and DEDUPE to the account. A single meeting ≠ a hot account; a cluster of decaying signals from multiple contacts = hot.\n- PQL (product-qualified) converts 5–6× MQL → prioritize product-usage signals when available.\n- App wiring: roll lead scores up by domain and multiply by an ICP-fit score derived from Company Profile / Brand Context.`,\n },\n {\n name: \"Unit economics that matter (2025–26 benchmarks)\",\n targets: [\"kpis\", \"recommendations\", \"attribution\"],\n content: `### Unit economics that matter\n- LTV:CAC ≥ 3 viable; top quartile 4–6×. CAC payback < 12–18 mo (PLG/SMB < 12). NRR ≥ 110–120% at scale. Rule of 40 (growth% + FCF margin%) ≥ 40, target 50+. Magic Number (net-new ARR / prior-Q S&M) > 0.75 = efficient.\n- LTV must use GROSS MARGIN and the retention curve, not ARPU × arbitrary months. CAC = fully-loaded acquisition spend ÷ new customers (not goal completions).\n- App wiring: compute from the Revenue sheet + spend; grade against these thresholds with healthy/borderline/unhealthy; flag any figure that is an estimate.`,\n },\n {\n name: \"Bullseye channel strategy\",\n targets: [\"recommendations\", \"chat\", \"attribution\"],\n content: `### Bullseye channel strategy\n- One channel DOMINATES acquisition at a time. Process: Brainstorm all 19 channels → Rank (inner/potential/longshot) → Prioritize 3 → Test cheaply (cost-per-acq? volume available? right customers?) → Focus on the winner; revisit when it saturates.\n- Don't diversify channels prematurely; diversify LOOPS, not unproven channels.\n- App wiring: a channel-test ledger (cost/volume/fit per channel from attribution) → recommend doubling the current inner-ring channel, testing the next 3, dropping longshots.`,\n },\n {\n name: \"Experiment rigor\",\n targets: [\"ice_suggest\", \"recommendations\"],\n content: `### Experiment rigor\n- Every test needs: a hypothesis (because X, if we do Y, metric Z moves by W), ONE success metric, a baseline, an MDE (minimum detectable effect), a sample size, and a SIGNIFICANCE call (don't read results before powered).\n- A result is a WIN only if significant; otherwise inconclusive. Learnings compound — feed realized win-rates back as priors so ICE confidence reflects evidence, not vibes.\n- Prefer high-ICE × addresses-the-binding-constraint; kill fast, document always.\n- App wiring: completeExperiment computes significance + writes the metric back via propagate; suggestICE uses win-rate-by-lever as its confidence prior.`,\n },\n];\n\nfunction estTokens(s: string): number { return Math.ceil((s || \"\").length / 4); }\n\nasync function getSeedVersion(): Promise<number> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = 'knowledge_seed_version'\");\n const raw = (r.rows?.[0] as any[])?.[0];\n if (raw == null) return 0;\n return parseInt(String(raw).replace(/^\"|\"$/g, \"\"), 10) || 0;\n } catch { return 0; }\n}\n\nasync function rowExists(name: string, source: string): Promise<boolean> {\n try {\n const r = await sqlite.execute(\"SELECT 1 FROM knowledge_files WHERE name = ? AND source = ? LIMIT 1\", [name, source]);\n return (r.rows?.length || 0) > 0;\n } catch { return false; }\n}\n\nasync function insertSeed(row: SeedRow): Promise<void> {\n // Idempotent: skip if a row with the same name+source already exists.\n if (await rowExists(row.name, row.source)) return;\n await sqlite.execute(\n `INSERT INTO knowledge_files (name, kind, content, summary, targets, source, tokens_estimate, enabled)\n VALUES (?, ?, ?, ?, ?, ?, ?, 1)`,\n [row.name, row.kind, row.content, row.content.slice(0, 1400), JSON.stringify(row.targets), row.source, estTokens(row.content)]\n );\n}\n\n// Inserts both packs once (gated by knowledge_seed_version). Safe to call on every boot.\nexport async function seedKnowledge(): Promise<{ seeded: number; skipped: boolean }> {\n try {\n const current = await getSeedVersion();\n if (current >= SEED_VERSION) return { seeded: 0, skipped: true };\n\n let seeded = 0;\n for (const m of MATH_KNOWLEDGE_SEEDS) {\n await insertSeed({ name: m.name, kind: m.kind, targets: m.targets, content: m.content, source: \"seed_math\" });\n seeded++;\n }\n for (const g of GROWTH_MODEL_SEEDS) {\n await insertSeed({ name: g.name, kind: \"growth-model\", targets: g.targets, content: g.content, source: \"seed_growth\" });\n seeded++;\n }\n\n await sqlite.execute(\n `INSERT INTO app_config (key, value) VALUES ('knowledge_seed_version', ?)\n ON CONFLICT(key) DO UPDATE SET value = excluded.value`,\n [String(SEED_VERSION)]\n );\n return { seeded, skipped: false };\n } catch (e) {\n // Never let seeding break boot.\n console.log(\"[knowledge_seed] failed:\", String((e as any)?.message || e));\n return { seeded: 0, skipped: true };\n }\n}\n",[],[365,795],"28442654-5d69-11f1-8587-1607ee4eb77e","knowledge.ts","692223bc-5989-11f1-aff7-ee650bb23af1",619,["D",1780284165312],"// Knowledge files — growth playbooks/frameworks AND math formulas. Each file targets one\n// or more AI features (chat, kpis, ice_suggest, funnel_summary, etc.) — when a feature\n// runs, only its targeted knowledge is injected. Empty targets = global (affects every\n// chat turn). Storage: knowledge_files table.\n//\n// Designed for token-efficient injection: AI summaries capped to ~500 tokens each; consumer\n// concatenates targeted+enabled summaries up to a per-call budget.\nimport { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\n// Registry of AI features that can be targeted by a knowledge file. Used by the graph UI\n// to render feature nodes and by chat/intelligence to look up applicable knowledge.\nexport const AI_FEATURES: Array<{ id: string; label: string; description: string }> = [\n { id: \"chat\", label: \"Chat Assistant\", description: \"General chat. Reads global + chat-targeted knowledge on every turn.\" },\n { id: \"funnel_summary\", label: \"Funnel Health\", description: \"TOFU/MOFU/BOFU stage scoring + overall composite.\" },\n { id: \"kpis\", label: \"AIO KPIs\", description: \"CAC, LTV, MER, Pipeline Coverage, funnel conversion.\" },\n { id: \"ice_suggest\", label: \"ICE Suggester\", description: \"LLM-assisted ICE/RICE/WSJF scoring for experiments.\" },\n { id: \"lead_scoring\", label: \"Lead Scoring\", description: \"MQL/SQL classification + intent signal weights.\" },\n { id: \"digest\", label: \"Weekly Digest\", description: \"Growth digest synthesis.\" },\n { id: \"standup\", label: \"Daily Standup\", description: \"Daily team brief.\" },\n { id: \"cohort_ltv\", label: \"Cohort LTV\", description: \"Cohort survival LTV computation.\" },\n { id: \"correlations\", label: \"Signal Correlations\", description: \"Cross-metric causal heuristics.\" },\n { id: \"recommendations\", label: \"Recommendations\", description: \"Top-3 next-best-action ranking.\" },\n];\nconst FEATURE_IDS = new Set(AI_FEATURES.map(f => f.id));\n\nexport interface KnowledgeFile {\n id: number;\n name: string;\n kind: string;\n content: string;\n summary: string | null;\n targets: string[]; // empty = global (affects every feature)\n source: string; // 'user' | 'seed_math'\n tokens_estimate: number;\n enabled: boolean;\n uploaded_at: string;\n updated_at: string;\n}\n\nfunction parseTargets(raw: any): string[] {\n if (Array.isArray(raw)) return raw.filter(t => typeof t === \"string\");\n if (typeof raw === \"string\" && raw.trim()) {\n try { const parsed = JSON.parse(raw); return Array.isArray(parsed) ? parsed.filter(t => typeof t === \"string\") : []; } catch { return []; }\n }\n return [];\n}\nfunction sanitizeTargets(targets: any): string[] {\n // Drop invalid feature ids (registry is the source of truth).\n return parseTargets(targets).filter(t => FEATURE_IDS.has(t));\n}\n\n// Rough chars/4 token estimator — good enough for prompt-budget caps without a tokenizer dep.\nconst estTokens = (s: string): number => Math.ceil((s || \"\").length / 4);\n\nfunction rowToFile(cols: string[], row: any[]): KnowledgeFile {\n const obj: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) obj[cols[i]] = row[i];\n return {\n id: Number(obj.id),\n name: String(obj.name),\n kind: String(obj.kind || \"playbook\"),\n content: String(obj.content || \"\"),\n summary: obj.summary ? String(obj.summary) : null,\n targets: parseTargets(obj.targets),\n source: String(obj.source || \"user\"),\n tokens_estimate: Number(obj.tokens_estimate || 0),\n enabled: Number(obj.enabled || 0) === 1,\n uploaded_at: String(obj.uploaded_at || \"\"),\n updated_at: String(obj.updated_at || \"\"),\n };\n}\n\n// ─── Listing — light (no content body) for the Sidebar / file list ──────────\nexport async function listKnowledgeFiles(): Promise<Array<Omit<KnowledgeFile, \"content\">>> {\n try {\n const r = await sqlite.execute(\n `SELECT id, name, kind, summary, targets, source, tokens_estimate, enabled, uploaded_at, updated_at\n FROM knowledge_files\n ORDER BY source = 'seed_math' DESC, uploaded_at DESC`\n );\n return (r.rows || []).map((row: any) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < r.columns.length; i++) obj[r.columns[i]] = row[i];\n return {\n id: Number(obj.id),\n name: String(obj.name),\n kind: String(obj.kind || \"playbook\"),\n summary: obj.summary ? String(obj.summary) : null,\n targets: parseTargets(obj.targets),\n source: String(obj.source || \"user\"),\n tokens_estimate: Number(obj.tokens_estimate || 0),\n enabled: Number(obj.enabled || 0) === 1,\n uploaded_at: String(obj.uploaded_at || \"\"),\n updated_at: String(obj.updated_at || \"\"),\n };\n });\n } catch {\n return [];\n }\n}\n\nexport async function getKnowledgeFile(id: string | number): Promise<KnowledgeFile | null> {\n try {\n const r = await sqlite.execute(\n `SELECT id, name, kind, content, summary, targets, source, tokens_estimate, enabled, uploaded_at, updated_at\n FROM knowledge_files WHERE id = ?`,\n [Number(id)]\n );\n if (!r.rows?.length) return null;\n return rowToFile(r.columns, r.rows[0] as any[]);\n } catch {\n return null;\n }\n}\n\n// ─── Create — accepts file content (already decoded). Generates AI summary opportunistically. ───\nexport async function createKnowledgeFile(data: {\n name: string;\n kind?: string;\n content: string;\n enabled?: boolean;\n targets?: string[];\n}): Promise<KnowledgeFile | null> {\n const name = (data.name || \"Untitled\").trim();\n const kind = (data.kind || \"playbook\").trim();\n const content = String(data.content || \"\");\n if (!content.trim()) return null;\n const tokens = estTokens(content);\n const enabled = data.enabled === false ? 0 : 1;\n const targets = JSON.stringify(sanitizeTargets(data.targets));\n\n await sqlite.execute(\n `INSERT INTO knowledge_files (name, kind, content, tokens_estimate, enabled, targets, source)\n VALUES (?, ?, ?, ?, ?, ?, 'user')`,\n [name, kind, content, tokens, enabled, targets]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = Number((idRow.rows[0] as any[])[0]);\n\n // Best-effort AI summary so future chats inject the gist (not the raw file).\n // We don't block the create call — fire-and-forget. If LLM fails, summary stays null.\n summarizeKnowledgeFile(id).catch(() => {});\n\n return await getKnowledgeFile(id);\n}\n\nexport async function updateKnowledgeFile(\n id: string | number,\n data: { name?: string; kind?: string; enabled?: boolean; content?: string; targets?: string[] }\n): Promise<KnowledgeFile | null> {\n const cur = await getKnowledgeFile(id);\n if (!cur) return null;\n const name = data.name?.trim() || cur.name;\n const kind = data.kind?.trim() || cur.kind;\n const enabled = typeof data.enabled === \"boolean\" ? (data.enabled ? 1 : 0) : (cur.enabled ? 1 : 0);\n const content = typeof data.content === \"string\" ? data.content : cur.content;\n const tokens = estTokens(content);\n const targets = JSON.stringify(\n Array.isArray(data.targets) ? sanitizeTargets(data.targets) : cur.targets\n );\n await sqlite.execute(\n `UPDATE knowledge_files\n SET name = ?, kind = ?, content = ?, tokens_estimate = ?, enabled = ?, targets = ?, updated_at = datetime('now')\n WHERE id = ?`,\n [name, kind, content, tokens, enabled, targets, Number(id)]\n );\n // Re-summarize when content changed.\n if (typeof data.content === \"string\" && data.content !== cur.content) {\n summarizeKnowledgeFile(id).catch(() => {});\n }\n return await getKnowledgeFile(id);\n}\n\nexport async function deleteKnowledgeFile(id: string | number): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM knowledge_files WHERE id = ?\", [Number(id)]);\n return true;\n } catch {\n return false;\n }\n}\n\n// ─── Summarization — LLM compresses content into a 300-500 token gist for chat injection ─\nexport async function summarizeKnowledgeFile(id: string | number): Promise<string | null> {\n const file = await getKnowledgeFile(id);\n if (!file || !file.content?.trim()) return null;\n\n const cfg = await getLLMConfig();\n if (!cfg.api_key) return null;\n\n // Short-circuit: if content is already <2000 chars, the content IS the summary.\n if (file.content.length < 2000) {\n const summary = file.content.trim();\n await sqlite.execute(\n \"UPDATE knowledge_files SET summary = ?, updated_at = datetime('now') WHERE id = ?\",\n [summary, Number(id)]\n );\n return summary;\n }\n\n const prompt = `You compress growth playbooks for an AI assistant. The assistant will read your\nsummary on every chat turn, so be dense and information-rich. Preserve:\n- Frameworks, principles, scoring rules, formulas\n- Concrete heuristics, numbers, thresholds\n- Decision rules (\"when X → do Y\")\n- Anti-patterns to avoid\nDrop: storytelling, examples that don't carry a rule, marketing fluff.\n\nOutput: 400-600 words. Markdown bullets. No preamble.\n\n---\nTITLE: ${file.name}\nKIND: ${file.kind}\n---\n${file.content.slice(0, 50000)}`;\n\n try {\n const resp = await callLLM(\n cfg,\n [{ role: \"user\", content: prompt }],\n { maxTokens: 1200 }\n );\n const summary = (resp.content || \"\").trim();\n if (!summary) return null;\n await sqlite.execute(\n \"UPDATE knowledge_files SET summary = ?, updated_at = datetime('now') WHERE id = ?\",\n [summary, Number(id)]\n );\n return summary;\n } catch {\n return null;\n }\n}\n\n// ─── Targeted injection helper — only loads knowledge for THIS feature ────────\n// Each feature can pull ONLY the knowledge files that listed it in `targets`, plus any\n// globally-targeted files (empty targets array). This keeps prompts lean per feature.\n//\n// Usage:\n// const ctx = await loadEnabledKnowledgeContext({ target: 'kpis', maxChars: 4000 });\n// // → returns markdown with only KPI-relevant playbooks\n//\n// Back-compat: calling with no target (or 'chat') returns global + chat-targeted files,\n// matching the previous behavior api/chat.ts depends on.\nexport async function loadEnabledKnowledgeContext(\n opts?: { target?: string; maxChars?: number } | number\n): Promise<string> {\n // Back-compat: prior signature was loadEnabledKnowledgeContext(maxChars)\n const maxChars = typeof opts === \"number\" ? opts : (opts?.maxChars ?? 6000);\n const target = typeof opts === \"object\" ? opts?.target : undefined;\n\n try {\n const r = await sqlite.execute(\n `SELECT name, kind, summary, content, targets\n FROM knowledge_files\n WHERE enabled = 1\n ORDER BY uploaded_at DESC`\n );\n const rows = (r.rows || []) as any[];\n if (rows.length === 0) return \"\";\n\n // Filter by target: include the file if its targets are EMPTY (global) or contain the requested target.\n const filtered = target\n ? rows.filter((row) => {\n const t = parseTargets(row[4]);\n return t.length === 0 || t.includes(target);\n })\n : rows;\n if (filtered.length === 0) return \"\";\n\n const header = target\n ? `## Active Growth Knowledge (filtered for \"${target}\")\\nApply the rules and formulas below when reasoning.\\n\\n`\n : \"## Active Growth Knowledge\\nThe user has uploaded these playbooks. Apply them when reasoning.\\n\\n\";\n let out = header;\n let remaining = maxChars - out.length;\n for (const row of filtered) {\n const name = String(row[0]);\n const kind = String(row[1] || \"playbook\");\n const summary = row[2] ? String(row[2]) : null;\n const content = row[3] ? String(row[3]) : \"\";\n const body = summary || content.slice(0, 800);\n const block = `### ${name} (${kind})\\n${body}\\n\\n`;\n if (block.length > remaining) {\n if (remaining > 200) out += block.slice(0, remaining - 12) + \"\\n…[trunc]\\n\";\n break;\n }\n out += block;\n remaining -= block.length;\n }\n return out.trim();\n } catch {\n return \"\";\n }\n}\n\n// ─── Knowledge graph builder — nodes + edges for the visual UI ───────────────\n// Used by GET /api/knowledge/graph. Returns:\n// - features: the AI_FEATURES registry (consumer reference)\n// - nodes: knowledge_files as 'knowledge' nodes + AI_FEATURES as 'feature' nodes\n// - edges: one edge per (file, target) pair. GLOBAL files (empty targets) get an\n// edge to every feature so the UI can render them as connected-to-all.\nexport async function buildKnowledgeGraph(): Promise<{\n features: typeof AI_FEATURES;\n nodes: Array<any>;\n edges: Array<{ from: number; to: string; global?: boolean }>;\n}> {\n const files = await listKnowledgeFiles();\n const featureNodes = AI_FEATURES.map((f) => ({\n id: \"feature:\" + f.id,\n type: \"feature\" as const,\n label: f.label,\n description: f.description,\n feature_id: f.id,\n }));\n const knowledgeNodes = files.map((f) => ({\n id: \"knowledge:\" + f.id,\n type: \"knowledge\" as const,\n label: f.name,\n knowledge_id: f.id,\n kind: f.kind,\n source: (f as any).source,\n enabled: f.enabled,\n targets: f.targets,\n tokens: f.tokens_estimate,\n }));\n const edges: Array<{ from: number; to: string; global?: boolean }> = [];\n for (const f of files) {\n if (!f.enabled) continue;\n if (f.targets.length === 0) {\n // GLOBAL — connect to every feature\n for (const feat of AI_FEATURES) {\n edges.push({ from: f.id, to: feat.id, global: true });\n }\n } else {\n for (const t of f.targets) {\n edges.push({ from: f.id, to: t });\n }\n }\n }\n return {\n features: AI_FEATURES,\n nodes: [...knowledgeNodes, ...featureNodes],\n edges,\n };\n}\n",[],[365,803],"27c0a900-5d69-11f1-ac6f-1607ee4eb77e","intelligence.ts","d5cf99da-580e-11f1-b0e3-ee650bb23af1",618,["D",1780284164483],12,"// Cross-Metric Correlation Engine + Unified Funnel + LLM-generated commentary feed.\n// Maps all 14 integrations to TOFU/MOFU/BOFU stages, fetches live metrics,\n// computes health scores against 30d baselines, surfaces causal correlations\n// with narrative + action prompts. Cached aggressively (5min TTL) since fetches\n// hit external APIs.\nimport { sqlite } from \"../db.ts\";\nimport { getCredentials, withGoogleFallback, safeJson } from \"./integrations/_shared.ts\";\nimport { getLLMConfig, callLLM } from \"./llm.ts\";\n\n// ─── FUNNEL MAPPING ─────────────────────────────────────────────────────────\ntype Stage = \"TOFU\" | \"MOFU\" | \"BOFU\";\n\nexport type AARRR = \"acquisition\" | \"activation\" | \"retention\" | \"referral\" | \"revenue\";\n\nexport interface MetricPoint {\n integration: string;\n metric: string;\n value: number;\n unit: string;\n stage: Stage;\n aarrr?: AARRR; // Pirate Metrics category\n recorded_at?: string; // ISO timestamp from metric_history\n}\n\n// Absolute health benchmarks per metric: maps raw value → 0-100 health.\n// Replaces the warm-start \"always 50%\" bug — gives real per-metric scoring based\n// on industry SaaS benchmarks. Returns null if no benchmark defined (falls back to\n// baseline-relative scoring).\nfunction normalizeAbsolute(integration: string, metric: string, value: number): number | null {\n const k = `${integration}:${metric}`;\n const clamp = (n: number) => Math.max(0, Math.min(100, n));\n switch (k) {\n // SEO authority (Ahrefs) — DR is already 0-100; refdomains log-scaled\n case \"ahrefs:domain_rating\": return clamp(value);\n case \"ahrefs:referring_domains\": return clamp(Math.log10(value + 1) * 25);\n // Search Console — CTR 0-5% maps to 0-100, position 1 = 100 falling 5%/rank\n case \"gsc:ctr\": return clamp((value / 5) * 100);\n case \"gsc:impressions\": return clamp(Math.log10(value + 1) * 20);\n case \"gsc:clicks\": return clamp(Math.log10(value + 1) * 25);\n case \"gsc:average_position\": return clamp(100 - (value - 1) * 5);\n // GA4 — engagement_time/180s = 100, conversions sqrt curve, bounce inverted\n case \"ga4:sessions\": return clamp(Math.log10(value + 1) * 20);\n case \"ga4:engagement_time\": return clamp((value / 180) * 100);\n case \"ga4:pages_per_session\": return clamp(((value - 1) / 3) * 100);\n case \"ga4:bounce_rate\": return clamp(100 - value);\n case \"ga4:conversions\": return clamp(Math.sqrt(value) * 15);\n // YouTube\n case \"youtube:views\": return clamp(Math.log10(value + 1) * 15);\n case \"youtube:subscribers\": return clamp(Math.log10(value + 1) * 20);\n // WordPress content velocity — 20 posts/period = full health\n case \"wordpress:published_posts\": return clamp(value * 5);\n // Communications volume — log-scaled\n case \"gmail:messages_total\": return clamp(Math.log10(value + 1) * 15);\n case \"gmail:threads_total\": return clamp(Math.log10(value + 1) * 18);\n case \"gmail:deals_initiated\": return clamp(value * 10);\n case \"outlook:inbox_count\": return clamp(Math.log10(value + 1) * 15);\n case \"outlook:meetings_booked\":return clamp(value * 10);\n // LinkedIn\n case \"linkedin:post_impressions\": return clamp(Math.log10(value + 1) * 15);\n default: return null; // no absolute benchmark → fall back to baseline-relative\n }\n}\n\n// AARRR mapping — same metric can belong to multiple categories.\n// gsc.average_position, gsc.ctr, youtube.subscribers, linkedin.reply_rate all carry\n// Pirate-Metric classification so the funnel summary never leaves stages undefined.\nconst AARRR_MAP: Record<string, AARRR> = {\n // Acquisition (first touch)\n \"gsc:clicks\": \"acquisition\", \"youtube:views\": \"acquisition\", \"linkedin:post_impressions\": \"acquisition\",\n \"gsc:impressions\": \"acquisition\", \"ahrefs:organic_keywords\": \"acquisition\",\n \"gsc:average_position\": \"acquisition\",\n // Activation (first meaningful engagement)\n \"ga4:sessions\": \"activation\", \"wordpress:published_posts\": \"activation\",\n \"gmail:messages_total\": \"activation\", \"gsc:ctr\": \"activation\",\n \"linkedin:reply_rate\": \"activation\",\n // Retention (repeated engagement)\n \"ga4:engagement_time\": \"retention\", \"ga4:pages_per_session\": \"retention\", \"ga4:bounce_rate\": \"retention\",\n \"gmail:threads_total\": \"retention\", \"youtube:subscribers\": \"retention\",\n // Referral (audience amplification)\n \"ahrefs:referring_domains\": \"referral\", \"ahrefs:domain_rating\": \"referral\",\n // Revenue (monetization events)\n \"ga4:conversions\": \"revenue\", \"outlook:inbox_count\": \"revenue\",\n \"outlook:meetings_booked\": \"revenue\", \"gmail:deals_initiated\": \"revenue\",\n};\nconst aarrrOf = (integration: string, metric: string): AARRR | undefined =>\n AARRR_MAP[`${integration}:${metric}`];\n\n// ─── BAYESIAN-INSPIRED CONFIDENCE ────────────────────────────────────────────\n// Penalize stale data, missing integrations, and rules currently inside their lag window.\nexport function computeConfidence(args: {\n source_freshness_days: number;\n target_freshness_days: number;\n source_has_data: boolean;\n target_has_data: boolean;\n lag_days: number;\n current_day_offset: number;\n}): number {\n let c = 1.0;\n if (args.source_freshness_days > 7) c *= 0.8;\n if (args.source_freshness_days > 30) c *= 0.6;\n if (args.target_freshness_days > 7) c *= 0.8;\n if (!args.source_has_data) c *= 0.3;\n if (!args.target_has_data) c *= 0.3;\n if (args.current_day_offset < args.lag_days) c *= 0.5;\n return Math.round(c * 100) / 100;\n}\n\n// ─── METRIC HISTORY (timeseries for delta_7d) ────────────────────────────────\nasync function recordMetric(integration: string, metric: string, value: number): Promise<void> {\n try {\n await sqlite.execute(\n \"INSERT INTO metric_history (integration, metric_key, value) VALUES (?, ?, ?)\",\n [integration, metric, value]\n );\n // Prune anything older than 60d to keep table bounded.\n await sqlite.execute(\n \"DELETE FROM metric_history WHERE recorded_at < datetime('now','-60 days')\"\n );\n } catch { /* non-critical */ }\n}\n\nasync function getMetricValueDaysAgo(integration: string, metric: string, days: number): Promise<number | null> {\n try {\n // Find closest recorded value to N days ago (within ±2 days window).\n const r = await sqlite.execute(\n `SELECT value FROM metric_history\n WHERE integration = ? AND metric_key = ?\n AND recorded_at <= datetime('now', ?)\n ORDER BY recorded_at DESC LIMIT 1`,\n [integration, metric, `-${days} days`]\n );\n if (!r.rows?.length) return null;\n return Number((r.rows[0] as any[])[0]);\n } catch { return null; }\n}\n\nexport interface CorrelationSignal {\n id: string;\n source_metric: { integration: string; metric: string; value: number; unit: string };\n target_metric: { integration: string; metric: string; value: number; unit: string };\n direction: \"positive\" | \"negative\" | \"neutral\";\n strength: \"strong\" | \"moderate\" | \"weak\";\n lag_days: number;\n narrative: string;\n funnel_stage_from: Stage;\n funnel_stage_to: Stage;\n confidence: number;\n action_prompt: string;\n}\n\nexport interface StageHealth {\n stage: Stage;\n label: string;\n description: string;\n metrics: Array<MetricPoint & { delta_pct: number | null; baseline: number | null; delta_7d?: number | null; delta_30d?: number | null }>;\n health_score: number; // 0-100\n diagnosis: string;\n}\n\nexport interface FunnelSummary {\n stages: StageHealth[];\n bottleneck: { stage: Stage; reason: string } | null;\n // Weighted composite — TOFU 0.25 · MOFU 0.40 · BOFU 0.35. BOFU revenue\n // matters more than top-of-funnel volume, MOFU drives qualified pipeline.\n overall_health: number;\n total_metrics_observed: number;\n total_integrations_connected: number;\n generated_at: string;\n}\n\nexport interface IntelligenceCard {\n type: \"correlation\" | \"anomaly\" | \"opportunity\" | \"risk\" | \"win\";\n priority: \"critical\" | \"high\" | \"medium\";\n title: string;\n body: string;\n metrics: string[];\n funnel_stages: Stage[];\n agent_name: string;\n timestamp: string;\n action_label: string;\n action_payload: { question: string; context?: any };\n}\n\n// ─── CACHE HELPERS ──────────────────────────────────────────────────────────\nasync function getCached(key: string): Promise<any | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, expires_at FROM intelligence_cache WHERE cache_key = ?\",\n [key]\n );\n if (!r.rows?.length) return null;\n const [payload, expiresAt] = r.rows[0] as any[];\n if (new Date(String(expiresAt)).getTime() < Date.now()) return null;\n return JSON.parse(String(payload));\n } catch { return null; }\n}\n\n// Read generated_at so the feed can invalidate itself when upstream summary is fresher.\nasync function getCachedTimestamp(key: string): Promise<{ payload: any; generated_at: number; expires_at: number } | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, generated_at, expires_at FROM intelligence_cache WHERE cache_key = ?\",\n [key]\n );\n if (!r.rows?.length) return null;\n const [payload, generatedAt, expiresAt] = r.rows[0] as any[];\n const exp = new Date(String(expiresAt)).getTime();\n if (exp < Date.now()) return null;\n return {\n payload: JSON.parse(String(payload)),\n generated_at: new Date(String(generatedAt)).getTime(),\n expires_at: exp,\n };\n } catch { return null; }\n}\n\nasync function setCached(key: string, payload: any, ttlSeconds: number): Promise<void> {\n const expiresAt = new Date(Date.now() + ttlSeconds * 1000).toISOString();\n await sqlite.execute(\n \"INSERT OR REPLACE INTO intelligence_cache (cache_key, payload, generated_at, expires_at) VALUES (?, ?, datetime('now'), ?)\",\n [key, JSON.stringify(payload), expiresAt]\n );\n}\n\nasync function getBaseline(integration: string, metricKey: string): Promise<number | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT baseline_value FROM metric_baselines WHERE integration = ? AND metric_key = ?\",\n [integration, metricKey]\n );\n if (!r.rows?.length) return null;\n return Number((r.rows[0] as any[])[0]);\n } catch { return null; }\n}\n\nasync function setBaseline(integration: string, metricKey: string, value: number): Promise<void> {\n await sqlite.execute(\n \"INSERT OR REPLACE INTO metric_baselines (integration, metric_key, baseline_value, last_updated) VALUES (?, ?, ?, datetime('now'))\",\n [integration, metricKey, value]\n );\n}\n\n// Rolling 30-day baseline. Replaces the frozen warm-start baseline once we have\n// at least 3 samples in metric_history, so health scoring drifts with reality instead\n// of forever comparing against the first observed value.\nasync function getRollingBaseline(integration: string, metricKey: string): Promise<{ avg: number; sample_count: number } | null> {\n try {\n const r = await sqlite.execute(\n `SELECT AVG(value) AS avg_v, COUNT(*) AS n\n FROM metric_history\n WHERE integration = ? AND metric_key = ?\n AND recorded_at >= datetime('now','-30 days')`,\n [integration, metricKey]\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any;\n const avg = Array.isArray(row) ? Number(row[0]) : Number(row.avg_v);\n const n = Array.isArray(row) ? Number(row[1]) : Number(row.n);\n if (!Number.isFinite(avg) || !Number.isFinite(n) || n < 3) return null;\n return { avg, sample_count: n };\n } catch { return null; }\n}\n\n// Persist the rolling figures back to metric_baselines so external readers see them too.\nasync function persistRolling(integration: string, metricKey: string, rolling: { avg: number; sample_count: number }): Promise<void> {\n try {\n await sqlite.execute(\n \"UPDATE metric_baselines SET rolling_avg = ?, sample_count = ?, last_updated = datetime('now') WHERE integration = ? AND metric_key = ?\",\n [rolling.avg, rolling.sample_count, integration, metricKey]\n );\n } catch { /* non-critical */ }\n}\n\n// ─── METRIC FETCHERS (live data from integrations) ──────────────────────────\n// Each returns MetricPoint[] or [] on error/missing creds. Never throws.\n\nasync function fetchAhrefs(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"ahrefs\");\n if (!creds?.api_key || !creds.extra_config?.target) return [];\n const target = creds.extra_config.target;\n const today = new Date().toISOString().slice(0, 10);\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(`https://api.ahrefs.com/v3/site-explorer/domain-rating?target=${encodeURIComponent(target)}&date=${today}`,\n { headers: { Authorization: `Bearer ${creds.api_key}` } });\n if (r.ok) {\n const d = await safeJson(r);\n const dr = d?.domain_rating?.domain_rating;\n if (typeof dr === \"number\") out.push({ integration: \"ahrefs\", metric: \"domain_rating\", value: dr, unit: \"DR\", stage: \"TOFU\" });\n }\n } catch {}\n try {\n const r = await fetch(`https://api.ahrefs.com/v3/site-explorer/backlinks-stats?target=${encodeURIComponent(target)}&date=${today}`,\n { headers: { Authorization: `Bearer ${creds.api_key}` } });\n if (r.ok) {\n const d = await safeJson(r);\n const rd = d?.metrics?.refdomains;\n if (typeof rd === \"number\") out.push({ integration: \"ahrefs\", metric: \"referring_domains\", value: rd, unit: \"domains\", stage: \"TOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchGSC(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"gsc\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token || !creds.extra_config?.site_url) return [];\n const siteUrl = encodeURIComponent(creds.extra_config.site_url);\n const end = new Date().toISOString().slice(0, 10);\n const start = new Date(Date.now() - days * 864e5).toISOString().slice(0, 10);\n try {\n const r = await fetch(`https://searchconsole.googleapis.com/webmasters/v3/sites/${siteUrl}/searchAnalytics/query`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({ startDate: start, endDate: end, dimensions: [], rowLimit: 1 }),\n });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const row = d?.rows?.[0];\n if (!row) return [];\n return [\n { integration: \"gsc\", metric: \"impressions\", value: row.impressions || 0, unit: \"impressions\", stage: \"TOFU\" },\n { integration: \"gsc\", metric: \"clicks\", value: row.clicks || 0, unit: \"clicks\", stage: \"TOFU\" },\n { integration: \"gsc\", metric: \"ctr\", value: (row.ctr || 0) * 100, unit: \"%\", stage: \"MOFU\" },\n { integration: \"gsc\", metric: \"average_position\", value: row.position || 0, unit: \"pos\", stage: \"TOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchGA4(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"ga4\");\n creds = await withGoogleFallback(creds ?? {});\n const propertyId = creds.extra_config?.property_id;\n if (!creds.access_token || !propertyId) return [];\n try {\n const r = await fetch(`https://analyticsdata.googleapis.com/v1beta/properties/${propertyId}:runReport`, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({\n metrics: [\n { name: \"sessions\" }, { name: \"userEngagementDuration\" }, { name: \"bounceRate\" },\n { name: \"conversions\" }, { name: \"screenPageViewsPerSession\" },\n ],\n dateRanges: [{ startDate: `${days}daysAgo`, endDate: \"today\" }],\n }),\n });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const row = d?.rows?.[0];\n if (!row) return [];\n const vals = row.metricValues || [];\n const num = (i: number) => parseFloat(vals[i]?.value || \"0\");\n const sessions = num(0);\n return [\n { integration: \"ga4\", metric: \"sessions\", value: sessions, unit: \"sessions\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"engagement_time\", value: sessions ? num(1) / sessions : 0, unit: \"sec/sess\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"bounce_rate\", value: num(2) * 100, unit: \"%\", stage: \"MOFU\" },\n { integration: \"ga4\", metric: \"conversions\", value: num(3), unit: \"conv\", stage: \"BOFU\" },\n { integration: \"ga4\", metric: \"pages_per_session\", value: num(4), unit: \"pages\", stage: \"MOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchYouTube(): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"youtube\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token && !creds.api_key) return [];\n const auth = creds.access_token ? { Authorization: `Bearer ${creds.access_token}` } : {};\n const keyParam = !creds.access_token && creds.api_key ? `&key=${encodeURIComponent(creds.api_key)}` : \"\";\n try {\n const url = creds.access_token\n ? `https://www.googleapis.com/youtube/v3/channels?part=statistics&mine=true`\n : `https://www.googleapis.com/youtube/v3/channels?part=statistics&chart=mostPopular&maxResults=1${keyParam}`;\n const r = await fetch(url, { headers: auth });\n if (!r.ok) return [];\n const d = await safeJson(r);\n const s = d?.items?.[0]?.statistics;\n if (!s) return [];\n return [\n { integration: \"youtube\", metric: \"views\", value: parseInt(s.viewCount || \"0\", 10), unit: \"views\", stage: \"TOFU\" },\n { integration: \"youtube\", metric: \"subscribers\", value: parseInt(s.subscriberCount || \"0\", 10), unit: \"subs\", stage: \"TOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchLinkedIn(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"linkedin\");\n if (!creds?.access_token) return [];\n try {\n const r = await fetch(\"https://api.linkedin.com/v2/me\", { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!r.ok) return [];\n return [\n { integration: \"linkedin\", metric: \"post_impressions\", value: 0, unit: \"impressions\", stage: \"TOFU\" },\n { integration: \"linkedin\", metric: \"reply_rate\", value: 0, unit: \"%\", stage: \"BOFU\" },\n ];\n } catch { return []; }\n}\n\nasync function fetchGmail(days = 30): Promise<MetricPoint[]> {\n let creds = await getCredentials(\"gmail\");\n creds = await withGoogleFallback(creds ?? {});\n if (!creds.access_token) return [];\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(\"https://gmail.googleapis.com/gmail/v1/users/me/profile\", {\n headers: { Authorization: `Bearer ${creds.access_token}` },\n });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"gmail\", metric: \"messages_total\", value: d?.messagesTotal || 0, unit: \"msgs\", stage: \"MOFU\" });\n out.push({ integration: \"gmail\", metric: \"threads_total\", value: d?.threadsTotal || 0, unit: \"threads\", stage: \"MOFU\" });\n }\n } catch {}\n // BOFU: deals_initiated — count Gmail threads with deal-intent keywords in the selected window.\n try {\n const q = encodeURIComponent(`newer_than:${days}d (pricing OR proposal OR contract OR quote OR \"sales agreement\")`);\n const r = await fetch(`https://gmail.googleapis.com/gmail/v1/users/me/messages?q=${q}&maxResults=1`,\n { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"gmail\", metric: \"deals_initiated\", value: d?.resultSizeEstimate || 0, unit: \"threads\", stage: \"BOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchOutlook(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"outlook\");\n if (!creds?.access_token) return [];\n const out: MetricPoint[] = [];\n try {\n const r = await fetch(\"https://graph.microsoft.com/v1.0/me/messages?$top=1&$count=true\", {\n headers: { Authorization: `Bearer ${creds.access_token}`, ConsistencyLevel: \"eventual\" },\n });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"outlook\", metric: \"inbox_count\", value: d?.[\"@odata.count\"] || 0, unit: \"msgs\", stage: \"MOFU\" });\n }\n } catch {}\n // BOFU: meetings booked (next 14 days) — real calendar query\n try {\n const start = new Date().toISOString();\n const end = new Date(Date.now() + 14 * 86400000).toISOString();\n const r = await fetch(`https://graph.microsoft.com/v1.0/me/calendar/events?$top=1&$count=true&$filter=start/dateTime ge '${start}' and start/dateTime le '${end}'`,\n { headers: { Authorization: `Bearer ${creds.access_token}`, ConsistencyLevel: \"eventual\" } });\n if (r.ok) {\n const d = await safeJson(r);\n out.push({ integration: \"outlook\", metric: \"meetings_booked\", value: d?.[\"@odata.count\"] || 0, unit: \"mtgs\", stage: \"BOFU\" });\n }\n } catch {}\n return out;\n}\n\nasync function fetchWordPress(): Promise<MetricPoint[]> {\n const creds = await getCredentials(\"wordpress\");\n const siteUrl = (creds?.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl || !creds?.api_key) return [];\n try {\n const auth = btoa(`${creds.api_key}:${creds.api_secret ?? \"\"}`);\n const r = await fetch(`${siteUrl}/wp-json/wp/v2/posts?per_page=1`, {\n headers: { Authorization: `Basic ${auth}` },\n });\n if (!r.ok) return [];\n const total = parseInt(r.headers.get(\"x-wp-total\") || \"0\", 10);\n return [{ integration: \"wordpress\", metric: \"published_posts\", value: total, unit: \"posts\", stage: \"MOFU\" }];\n } catch { return []; }\n}\n\n// ─── FUNNEL SUMMARY ─────────────────────────────────────────────────────────\nconst STAGE_META: Record<Stage, { label: string; description: string }> = {\n TOFU: { label: \"Awareness & Reach\", description: \"Top-of-funnel visibility — SEO, social, content distribution.\" },\n MOFU: { label: \"Engagement & Consideration\", description: \"Mid-funnel signal — site engagement, content depth, outreach reply.\" },\n BOFU: { label: \"Conversion & Revenue\", description: \"Bottom-of-funnel — goal completions, deals, meetings, revenue.\" },\n};\n\nexport async function getFunnelSummary(force = false, days = 30): Promise<FunnelSummary> {\n const cacheKey = `funnel_summary_${days}d`;\n if (!force) {\n const cached = await getCached(cacheKey);\n if (cached) return cached;\n }\n\n const results = await Promise.allSettled([\n fetchAhrefs(), fetchGSC(days), fetchGA4(days), fetchYouTube(),\n fetchLinkedIn(), fetchGmail(days), fetchOutlook(), fetchWordPress(),\n ]);\n const allMetrics: MetricPoint[] = results\n .filter((r): r is PromiseFulfilledResult<MetricPoint[]> => r.status === \"fulfilled\")\n .flatMap((r) => r.value)\n .map((m) => ({ ...m, aarrr: aarrrOf(m.integration, m.metric), recorded_at: new Date().toISOString() }));\n\n // Update baselines on first run + write every value to metric_history for delta_7d\n for (const m of allMetrics) {\n const b = await getBaseline(m.integration, m.metric);\n if (b === null) await setBaseline(m.integration, m.metric, m.value);\n await recordMetric(m.integration, m.metric, m.value);\n }\n\n // Compute per-stage health. Each metric gets baseline delta + 7-day delta from metric_history.\n const stages: StageHealth[] = ([\"TOFU\", \"MOFU\", \"BOFU\"] as Stage[]).map((stage) => {\n const stageMetrics = allMetrics.filter((m) => m.stage === stage);\n const withDeltas = stageMetrics.map(async (m) => {\n // Prefer rolling 30d baseline once we have ≥3 samples; fall back to frozen warm-start.\n const frozen = await getBaseline(m.integration, m.metric);\n const rolling = await getRollingBaseline(m.integration, m.metric);\n const baseline = rolling ? rolling.avg : frozen;\n if (rolling) await persistRolling(m.integration, m.metric, rolling);\n const delta_pct = baseline && baseline !== 0 ? ((m.value - baseline) / Math.abs(baseline)) * 100 : null;\n // delta_7d / delta_30d: real timeseries comparison vs same metric N days ago (A2).\n // TOFU stage UI prefers 30d (SEO/branding moves slowly); MOFU/BOFU keep 7d for tighter feedback.\n const [v7, v30] = await Promise.all([\n getMetricValueDaysAgo(m.integration, m.metric, 7),\n getMetricValueDaysAgo(m.integration, m.metric, 30),\n ]);\n const delta_7d = v7 != null && v7 !== 0 ? ((m.value - v7) / Math.abs(v7)) * 100 : null;\n const delta_30d = v30 != null && v30 !== 0 ? ((m.value - v30) / Math.abs(v30)) * 100 : null;\n return { ...m, delta_pct, baseline, delta_7d, delta_30d };\n });\n return { stage, metrics: withDeltas, stageMetrics };\n }) as any;\n\n // Resolve the promises in metrics.\n const resolvedStages: StageHealth[] = await Promise.all(\n stages.map(async (s: any) => {\n const metrics = await Promise.all(s.metrics);\n if (metrics.length === 0) {\n return {\n stage: s.stage,\n label: STAGE_META[s.stage as Stage].label,\n description: STAGE_META[s.stage as Stage].description,\n metrics,\n health_score: 0,\n diagnosis: \"No data — connect integrations for this stage.\",\n };\n }\n\n // ── ABSOLUTE benchmark-based health (replaces broken warm-start baseline scoring) ──\n // Each metric gets a 0-100 score against industry benchmarks. Stage health = mean.\n const absoluteScores: number[] = [];\n const deltaContributions: number[] = [];\n for (const m of metrics as any[]) {\n const abs = normalizeAbsolute(m.integration, m.metric, m.value);\n if (abs !== null) absoluteScores.push(abs);\n // Chronological delta_7d bonus: ±10 added to absolute score if trending.\n if (m.delta_7d != null && abs !== null) {\n const BAD_WHEN_UP = new Set([\"bounce_rate\"]);\n const direction = BAD_WHEN_UP.has(m.metric) ? -1 : 1;\n deltaContributions.push(Math.max(-15, Math.min(15, m.delta_7d * direction * 0.5)));\n }\n }\n const baseHealth = absoluteScores.length\n ? absoluteScores.reduce((a, b) => a + b, 0) / absoluteScores.length\n : 50;\n const deltaBonus = deltaContributions.length\n ? deltaContributions.reduce((a, b) => a + b, 0) / deltaContributions.length\n : 0;\n const health_score = Math.max(0, Math.min(100, Math.round(baseHealth + deltaBonus)));\n\n // Chronological-aware diagnosis: state of absolute health + 7d trend direction.\n const avgTrend = deltaContributions.length\n ? deltaContributions.reduce((a, b) => a + b, 0) / deltaContributions.length\n : 0;\n let diagnosis: string;\n if (health_score >= 75) diagnosis = \"Healthy — metrics meet or exceed SaaS benchmarks.\" + (avgTrend > 2 ? \" Trending up week-over-week.\" : avgTrend < -2 ? \" Mild 7d softening — watch closely.\" : \"\");\n else if (health_score >= 50) diagnosis = \"Moderate — room to improve against benchmarks.\" + (avgTrend > 2 ? \" 7d momentum is positive.\" : avgTrend < -2 ? \" 7d trend is negative.\" : \"\");\n else if (health_score >= 25) diagnosis = \"Underperforming benchmarks — focus optimization here.\" + (avgTrend < -2 ? \" Compounding 7d decline.\" : \"\");\n else diagnosis = \"Critical — metrics far below industry baselines. Prioritize this stage.\";\n\n return {\n stage: s.stage,\n label: STAGE_META[s.stage as Stage].label,\n description: STAGE_META[s.stage as Stage].description,\n metrics,\n health_score,\n diagnosis,\n };\n })\n );\n\n // Bottleneck = lowest-health stage with data.\n const candidates = resolvedStages.filter((s) => s.metrics.length > 0);\n const bottleneckStage = candidates.length\n ? candidates.reduce((lo, s) => (s.health_score < lo.health_score ? s : lo))\n : null;\n\n // Weighted overall_health — only counts stages with data so a missing integration\n // doesn't crater the composite. Weights renormalize when a stage is missing.\n const stageWeights: Record<Stage, number> = { TOFU: 0.25, MOFU: 0.40, BOFU: 0.35 };\n const dataStages = resolvedStages.filter((s) => s.metrics.length > 0);\n const overallHealth = dataStages.length\n ? (() => {\n const weightSum = dataStages.reduce((acc, s) => acc + stageWeights[s.stage], 0);\n const weighted = dataStages.reduce((acc, s) => acc + s.health_score * stageWeights[s.stage], 0);\n return Math.max(0, Math.min(100, weighted / weightSum));\n })()\n : 0;\n\n const summary: FunnelSummary = {\n stages: resolvedStages,\n bottleneck: bottleneckStage\n ? { stage: bottleneckStage.stage, reason: bottleneckStage.diagnosis }\n : null,\n overall_health: Math.round(overallHealth),\n total_metrics_observed: allMetrics.length,\n total_integrations_connected: new Set(allMetrics.map((m) => m.integration)).size,\n generated_at: new Date().toISOString(),\n };\n\n await setCached(cacheKey, summary, 300);\n // Also cache under the legacy key so getCorrelations() (which calls getFunnelSummary() without\n // days) can still find a warm entry when the default 30d summary was just computed.\n if (days === 30) await setCached(\"funnel_summary\", summary, 300);\n return summary;\n}\n\n// ─── CORRELATIONS ────────────────────────────────────────────────────────────\nfunction getMetric(all: MetricPoint[], integration: string, metric: string): MetricPoint | undefined {\n return all.find((m) => m.integration === integration && m.metric === metric);\n}\n\nexport async function getCorrelations(force = false): Promise<CorrelationSignal[]> {\n if (!force) {\n const cached = await getCached(\"correlations\");\n if (cached) return cached;\n }\n const summary = await getFunnelSummary();\n const all = summary.stages.flatMap((s) => s.metrics);\n const signals: CorrelationSignal[] = [];\n\n const dr = getMetric(all, \"ahrefs\", \"domain_rating\");\n const gscClicks = getMetric(all, \"gsc\", \"clicks\");\n const gscCtr = getMetric(all, \"gsc\", \"ctr\");\n const gscImp = getMetric(all, \"gsc\", \"impressions\");\n const sessions = getMetric(all, \"ga4\", \"sessions\");\n const engTime = getMetric(all, \"ga4\", \"engagement_time\");\n const conv = getMetric(all, \"ga4\", \"conversions\");\n const bounce = getMetric(all, \"ga4\", \"bounce_rate\");\n const liImp = getMetric(all, \"linkedin\", \"post_impressions\");\n\n const deltaOf = (m: any) => (m?.delta_pct ?? 0) as number;\n\n // Rule 1: Ahrefs DR up + GSC clicks up\n if (dr && gscClicks && deltaOf(dr) > 0 && deltaOf(gscClicks) > 0) {\n signals.push({\n id: \"dr-clicks\", source_metric: { integration: dr.integration, metric: dr.metric, value: dr.value, unit: dr.unit },\n target_metric: { integration: gscClicks.integration, metric: gscClicks.metric, value: gscClicks.value, unit: gscClicks.unit },\n direction: \"positive\", strength: \"strong\", lag_days: 30,\n narrative: \"Domain authority growth is converting to organic search clicks — SEO investment is compounding.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.78,\n action_prompt: \"Double down on the keyword clusters where rankings improved this month — let the AI prioritize.\",\n });\n }\n // Rule 2: GSC CTR up + sessions up\n if (gscCtr && sessions && deltaOf(gscCtr) > 0 && deltaOf(sessions) > 0) {\n signals.push({\n id: \"ctr-sessions\", source_metric: { integration: gscCtr.integration, metric: gscCtr.metric, value: gscCtr.value, unit: gscCtr.unit },\n target_metric: { integration: sessions.integration, metric: sessions.metric, value: sessions.value, unit: sessions.unit },\n direction: \"positive\", strength: \"strong\", lag_days: 7,\n narrative: \"Search result click-through improvements are driving qualified session growth.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.82,\n action_prompt: \"Identify the top 5 pages whose CTR improved and roll their title/meta patterns to similar pages.\",\n });\n }\n // Rule 3: engagement time up + conversions up\n if (engTime && conv && deltaOf(engTime) > 0 && deltaOf(conv) > 0) {\n signals.push({\n id: \"eng-conv\", source_metric: { integration: engTime.integration, metric: engTime.metric, value: engTime.value, unit: engTime.unit },\n target_metric: { integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit },\n direction: \"positive\", strength: \"moderate\", lag_days: 14,\n narrative: \"Content quality improvements correlate with downstream conversion lift.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.68,\n action_prompt: \"Pin the highest-engagement pages above the fold in nav — they're proven to convert.\",\n });\n }\n // Rule 4: high impressions + low CTR\n if (gscImp && gscCtr && gscImp.value > 1000 && gscCtr.value < 3) {\n signals.push({\n id: \"imp-low-ctr\", source_metric: { integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit },\n target_metric: { integration: gscCtr.integration, metric: gscCtr.metric, value: gscCtr.value, unit: gscCtr.unit },\n direction: \"negative\", strength: \"moderate\", lag_days: 0,\n narrative: \"High impressions with poor CTR indicates title/meta description optimization opportunity.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.85,\n action_prompt: \"Run a Content Weaver agent to rewrite titles + metas for the top 20 high-impression / low-CTR queries.\",\n });\n }\n // Rule 5: bounce up + conversions down (funnel leak)\n if (bounce && conv && deltaOf(bounce) > 5 && deltaOf(conv) < -5) {\n signals.push({\n id: \"bounce-conv-leak\", source_metric: { integration: bounce.integration, metric: bounce.metric, value: bounce.value, unit: bounce.unit },\n target_metric: { integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit },\n direction: \"negative\", strength: \"strong\", lag_days: 0,\n narrative: \"Traffic quality or landing page relevance degrading — funnel leakage at MOFU.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.74,\n action_prompt: \"Pull the top entry pages with rising bounce, audit message-market fit + page speed.\",\n });\n }\n // Rule 6: LinkedIn impressions up — social priming\n if (liImp && deltaOf(liImp) > 10) {\n signals.push({\n id: \"li-warmup\", source_metric: { integration: liImp.integration, metric: liImp.metric, value: liImp.value, unit: liImp.unit },\n target_metric: { integration: \"gmail\", metric: \"open_rate\", value: 0, unit: \"%\" },\n direction: \"positive\", strength: \"moderate\", lag_days: 21,\n narrative: \"Brand visibility on LinkedIn is warming cold email outreach — social priming effect.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"MOFU\", confidence: 0.55,\n action_prompt: \"Schedule outreach sequences to land 3-5 days after high-engagement LinkedIn posts.\",\n });\n }\n\n // Additional correlation rules (extends set from 6 → 12).\n const refDom = getMetric(all, \"ahrefs\", \"referring_domains\");\n const posts = getMetric(all, \"wordpress\", \"published_posts\");\n const pps = getMetric(all, \"ga4\", \"pages_per_session\");\n\n // Rule 7: DR ↑ → referring_domains ↑ (lag 14-30d, moderate)\n if (dr && refDom && deltaOf(dr) > 0 && deltaOf(refDom) > 0) {\n signals.push({\n id: \"dr-refdomains\", source_metric: {integration: dr.integration, metric: dr.metric, value: dr.value, unit: dr.unit},\n target_metric: {integration: refDom.integration, metric: refDom.metric, value: refDom.value, unit: refDom.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 21,\n narrative: \"Higher domain authority is attracting new backlinks as more sites notice your ranked content.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.65,\n action_prompt: \"Identify the top 3 pages gaining new backlinks and double down on their content angle.\",\n });\n }\n // Rule 8: referring_domains ↑ → GSC impressions ↑ (lag 14-21d, strong)\n if (refDom && gscImp && deltaOf(refDom) > 0 && deltaOf(gscImp) > 0) {\n signals.push({\n id: \"refdomains-impressions\", source_metric: {integration: refDom.integration, metric: refDom.metric, value: refDom.value, unit: refDom.unit},\n target_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n direction: \"positive\", strength: \"strong\", lag_days: 18,\n narrative: \"New backlinks are pushing more keywords into ranking range, expanding search visibility.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.72,\n action_prompt: \"Pitch guest posts to high-DR sites in the same vertical to amplify the lift.\",\n });\n }\n // Rule 9: GSC impressions ↑ + CTR stable → GSC clicks ↑ (lag 0-7d, strong, near 1:1)\n if (gscImp && gscClicks && deltaOf(gscImp) > 5 && Math.abs(deltaOf(gscCtr) || 0) < 5 && deltaOf(gscClicks) > 0) {\n signals.push({\n id: \"imp-clicks\", source_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n target_metric: {integration: gscClicks.integration, metric: gscClicks.metric, value: gscClicks.value, unit: gscClicks.unit},\n direction: \"positive\", strength: \"strong\", lag_days: 3,\n narrative: \"Impression growth with stable CTR is converting directly to click volume — pure ranking lift.\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.80,\n action_prompt: \"Track which queries gained impressions this period and ensure landing pages can absorb the traffic.\",\n });\n }\n // Rule 10: WordPress posts ↑ → GSC impressions ↑ (lag 30-90d, moderate)\n if (posts && gscImp && deltaOf(posts) > 0 && deltaOf(gscImp) > 0) {\n signals.push({\n id: \"posts-impressions\", source_metric: {integration: posts.integration, metric: posts.metric, value: posts.value, unit: posts.unit},\n target_metric: {integration: gscImp.integration, metric: gscImp.metric, value: gscImp.value, unit: gscImp.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 60,\n narrative: \"Recent content publishing cadence is now accumulating ranking signals (4-12 week lag).\",\n funnel_stage_from: \"TOFU\", funnel_stage_to: \"TOFU\", confidence: 0.58,\n action_prompt: \"Maintain publishing cadence — backlog of posts will continue to compound impressions over 90 days.\",\n });\n }\n // Rule 11: pages_per_session ↑ → conversions ↑ (lag 7d, moderate)\n if (pps && conv && deltaOf(pps) > 0 && deltaOf(conv) > 0) {\n signals.push({\n id: \"pps-cvr\", source_metric: {integration: pps.integration, metric: pps.metric, value: pps.value, unit: pps.unit},\n target_metric: {integration: conv.integration, metric: conv.metric, value: conv.value, unit: conv.unit},\n direction: \"positive\", strength: \"moderate\", lag_days: 7,\n narrative: \"Deeper site exploration correlates with higher purchase intent — content map is working.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.62,\n action_prompt: \"Surface high-engagement secondary pages in primary nav to push more visitors into the explore loop.\",\n });\n }\n // Rule 12: Gmail messages_total trend — proxy for outreach velocity (qualified-interest signal)\n const gmailMsgs = getMetric(all, \"gmail\", \"messages_total\");\n if (gmailMsgs && deltaOf(gmailMsgs) > 10) {\n signals.push({\n id: \"gmail-velocity\", source_metric: {integration: gmailMsgs.integration, metric: gmailMsgs.metric, value: gmailMsgs.value, unit: gmailMsgs.unit},\n target_metric: {integration: \"outlook\", metric: \"meetings_booked\", value: 0, unit: \"mtgs\"},\n direction: \"positive\", strength: \"moderate\", lag_days: 3,\n narrative: \"Outbound email volume is up — reply→meeting conversion typically follows within 2-5 days.\",\n funnel_stage_from: \"MOFU\", funnel_stage_to: \"BOFU\", confidence: 0.50,\n action_prompt: \"Block calendar windows next week — meeting requests usually cluster 3-5 days after email volume spikes.\",\n });\n }\n\n await setCached(\"correlations\", signals, 300);\n return signals;\n}\n\n// ─── RECOMMENDATIONS ────────────────────────────────────────────────────────\nexport async function getRecommendations(): Promise<Array<{ title: string; action: string; priority: string }>> {\n const cached = await getCached(\"recommendations\");\n if (cached) return cached;\n\n const summary = await getFunnelSummary();\n const correlations = await getCorrelations();\n const recs: Array<{ title: string; action: string; priority: string }> = [];\n\n // Top recommendation: address bottleneck\n if (summary.bottleneck) {\n recs.push({\n title: `Fix the ${summary.bottleneck.stage} bottleneck`,\n action: `Why is ${summary.bottleneck.stage} health low?`,\n priority: \"critical\",\n });\n }\n // Highest-confidence positive signal: amplify it\n const positives = correlations.filter((c) => c.direction === \"positive\").sort((a, b) => b.confidence - a.confidence);\n if (positives[0]) {\n recs.push({\n title: positives[0].narrative.slice(0, 60) + \"…\",\n action: `How do I amplify the ${positives[0].source_metric.metric} → ${positives[0].target_metric.metric} chain?`,\n priority: \"high\",\n });\n }\n // Highest-confidence negative signal: investigate\n const negatives = correlations.filter((c) => c.direction === \"negative\").sort((a, b) => b.confidence - a.confidence);\n if (negatives[0]) {\n recs.push({\n title: `Investigate: ${negatives[0].narrative.slice(0, 50)}…`,\n action: negatives[0].action_prompt,\n priority: \"high\",\n });\n }\n\n // Fallback if no real data\n if (recs.length === 0) {\n recs.push(\n { title: \"Connect Google Workspace to unlock funnel intelligence\", action: \"How do I connect Google?\", priority: \"high\" },\n { title: \"Connect Ahrefs to track TOFU authority signals\", action: \"What does Ahrefs add to my funnel?\", priority: \"medium\" },\n { title: \"Configure GA4 property ID for MOFU/BOFU visibility\", action: \"How do I set up GA4?\", priority: \"medium\" },\n );\n }\n\n await setCached(\"recommendations\", recs.slice(0, 5), 300);\n return recs.slice(0, 5);\n}\n\n// ─── MQL / SQL LEAD SCORING ──────────────────────────────────────────────────\n// B2B SaaS qualification math:\n// MQL score = sum of behavioral signals (opens, clicks, pricing views, downloads)\n// MQL threshold: ≥ 40\n// SQL = 2+ qualified signals (reply, accepted connection, booked meeting, requested demo)\n// Since Nexus has no CRM, we derive from Gmail inbox threads + Outlook calendar overlap.\nexport interface LeadScore {\n email: string;\n name: string;\n company: string;\n mql_score: number;\n mql_qualified: boolean; // score >= 40\n sql_signals: string[]; // list of qualified signals fired\n sql_qualified: boolean; // signals.length >= 2\n stage: \"cold\" | \"MQL\" | \"SQL\";\n signal_breakdown: Record<string, number>;\n source: \"gmail\" | \"outlook\" | \"linkedin\";\n last_touch_at?: string;\n}\n\nexport async function getLeadScores(): Promise<LeadScore[]> {\n const cached = await getCached(\"lead_scores\");\n if (cached) return cached;\n\n const leads = new Map<string, LeadScore>();\n const getLead = (email: string, name: string, company: string, source: any): LeadScore => {\n if (!leads.has(email)) {\n leads.set(email, {\n email, name, company, source,\n mql_score: 0, mql_qualified: false,\n sql_signals: [], sql_qualified: false,\n stage: \"cold\", signal_breakdown: {},\n });\n }\n return leads.get(email)!;\n };\n\n // ── Pull Outlook inbox: each sender gets points for engagement signals ──\n let outlookCreds = await getCredentials(\"outlook\");\n if (outlookCreds?.access_token) {\n try {\n const r = await fetch(\"https://graph.microsoft.com/v1.0/me/messages?$top=50&$orderby=receivedDateTime desc&$select=from,subject,isRead,body,receivedDateTime\",\n { headers: { Authorization: `Bearer ${outlookCreds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n for (const m of (d.value || [])) {\n const email = m.from?.emailAddress?.address;\n if (!email) continue;\n const name = m.from?.emailAddress?.name || email.split(\"@\")[0];\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = getLead(email, name, company, \"outlook\");\n const subj = (m.subject || \"\").toLowerCase();\n const body = (m.body?.content || \"\").toLowerCase();\n // MQL signals\n if (m.isRead === false) { lead.mql_score += 5; lead.signal_breakdown.unread = (lead.signal_breakdown.unread||0)+5; }\n if (/click|visited|opened/.test(body)) { lead.mql_score += 15; lead.signal_breakdown.clicks_link = 15; }\n if (/pricing|plans|cost/.test(subj+body)) { lead.mql_score += 25; lead.signal_breakdown.visits_pricing = 25; }\n if (/download|whitepaper|ebook|guide/.test(subj+body)) { lead.mql_score += 20; lead.signal_breakdown.downloads_content = 20; }\n if (/webinar|registered/.test(subj+body)) { lead.mql_score += 20; lead.signal_breakdown.attends_webinar = 20; }\n lead.mql_score += 10; lead.signal_breakdown.opens_email_sequence = 10;\n // SQL signals\n if (/re:|reply|response/i.test(subj) && !subj.startsWith(\"re:\")) {\n if (!lead.sql_signals.includes(\"reply_to_outreach_email\")) lead.sql_signals.push(\"reply_to_outreach_email\");\n }\n if (/demo|trial|proposal|quote/.test(subj+body)) {\n if (!lead.sql_signals.includes(\"requested_demo_or_pricing\")) lead.sql_signals.push(\"requested_demo_or_pricing\");\n }\n lead.last_touch_at = m.receivedDateTime || lead.last_touch_at;\n }\n }\n } catch {}\n\n // ── Pull Outlook calendar to detect booked discovery calls ──\n try {\n const start = new Date().toISOString();\n const end = new Date(Date.now() + 30 * 86400000).toISOString();\n const r = await fetch(`https://graph.microsoft.com/v1.0/me/calendar/events?$top=50&$filter=start/dateTime ge '${start}' and start/dateTime le '${end}'&$select=attendees,subject`,\n { headers: { Authorization: `Bearer ${outlookCreds.access_token}` } });\n if (r.ok) {\n const d = await safeJson(r);\n for (const ev of (d.value || [])) {\n for (const a of (ev.attendees || [])) {\n const email = a.emailAddress?.address;\n if (!email) continue;\n const name = a.emailAddress?.name || email.split(\"@\")[0];\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = leads.get(email) || getLead(email, name, company, \"outlook\");\n if (!lead.sql_signals.includes(\"booked_discovery_call\")) lead.sql_signals.push(\"booked_discovery_call\");\n }\n }\n }\n } catch {}\n }\n\n // ── (A6) Gmail thread analysis — runs independent of Outlook so Google-only users still get leads.\n // Mirrors Outlook deal-intent scoring against the recent inbox.\n try {\n const gmailCreds = await getCredentials(\"gmail\");\n const token = gmailCreds?.access_token;\n if (token) {\n const listR = await fetch(\"https://gmail.googleapis.com/gmail/v1/users/me/messages?maxResults=30&q=in:inbox+newer_than:30d\",\n { headers: { Authorization: `Bearer ${token}` } });\n if (listR.ok) {\n const list = await safeJson(listR);\n const ids: string[] = (list.messages || []).map((m: any) => m.id).slice(0, 30);\n const fetched = await Promise.allSettled(ids.map((id) =>\n fetch(`https://gmail.googleapis.com/gmail/v1/users/me/messages/${id}?format=metadata&metadataHeaders=From&metadataHeaders=Subject&metadataHeaders=Date`,\n { headers: { Authorization: `Bearer ${token}` } }).then((r) => r.ok ? r.json() : null)\n ));\n for (const f of fetched) {\n if (f.status !== \"fulfilled\" || !f.value) continue;\n const msg: any = f.value;\n const headers: any[] = msg.payload?.headers || [];\n const findH = (n: string) => headers.find((h) => h.name?.toLowerCase() === n.toLowerCase())?.value || \"\";\n const from = findH(\"From\");\n const subject = findH(\"Subject\");\n const date = findH(\"Date\");\n const snippet = (msg.snippet || \"\").toLowerCase();\n const emailMatch = from.match(/<([^>]+)>|([\\w.+-]+@[\\w.-]+)/);\n const email = (emailMatch?.[1] || emailMatch?.[2] || \"\").toLowerCase();\n if (!email) continue;\n const nameMatch = from.match(/^([^<]+)</);\n const name = (nameMatch?.[1] || email.split(\"@\")[0]).trim().replace(/\"/g, \"\");\n const company = (email.split(\"@\")[1] || \"\").split(\".\")[0].toUpperCase();\n const lead = leads.get(email) || getLead(email, name, company, \"gmail\");\n const subj = (subject || \"\").toLowerCase();\n const text = subj + \" \" + snippet;\n // Deal-intent MQL scoring weights\n if (/pricing|plans|cost|how much/.test(text)) {\n lead.mql_score += 25; lead.signal_breakdown.gmail_pricing = (lead.signal_breakdown.gmail_pricing || 0) + 25;\n }\n if (/proposal|quote|contract|sow|statement of work/.test(text)) {\n lead.mql_score += 20; lead.signal_breakdown.gmail_proposal = (lead.signal_breakdown.gmail_proposal || 0) + 20;\n if (!lead.sql_signals.includes(\"requested_proposal\")) lead.sql_signals.push(\"requested_proposal\");\n }\n if (/demo|trial|walkthrough|kickoff/.test(text)) {\n lead.mql_score += 20; lead.signal_breakdown.gmail_demo = (lead.signal_breakdown.gmail_demo || 0) + 20;\n if (!lead.sql_signals.includes(\"requested_demo_or_trial\")) lead.sql_signals.push(\"requested_demo_or_trial\");\n }\n if (/^re:/i.test(subj.trim())) {\n if (!lead.sql_signals.includes(\"replied_to_thread\")) lead.sql_signals.push(\"replied_to_thread\");\n }\n lead.mql_score += 5; lead.signal_breakdown.gmail_touch = (lead.signal_breakdown.gmail_touch || 0) + 5;\n if (date) {\n const dt = new Date(date).toISOString();\n if (!lead.last_touch_at || dt > lead.last_touch_at) lead.last_touch_at = dt;\n }\n }\n }\n }\n } catch { /* gmail optional */ }\n\n // Finalize: classify each lead\n const result: LeadScore[] = [];\n for (const lead of leads.values()) {\n lead.mql_qualified = lead.mql_score >= 40;\n lead.sql_qualified = lead.sql_signals.length >= 2;\n lead.stage = lead.sql_qualified ? \"SQL\" : lead.mql_qualified ? \"MQL\" : \"cold\";\n result.push(lead);\n }\n // Record state transitions so engagement velocity has data to compute from.\n // recordLeadStateChange skips inserts when the state hasn't changed since last write.\n try {\n const { recordLeadStateChange } = await import(\"./aio.ts\");\n await Promise.allSettled(result.map((l) => recordLeadStateChange(l.email, l.stage)));\n } catch { /* non-critical */ }\n // Sort: SQL first, then MQL, then by score desc\n result.sort((a, b) => {\n const rank = (l: LeadScore) => l.sql_qualified ? 2 : l.mql_qualified ? 1 : 0;\n return (rank(b) - rank(a)) || (b.mql_score - a.mql_score);\n });\n\n const sliced = result.slice(0, 50);\n await setCached(\"lead_scores\", sliced, 300);\n return sliced;\n}\n\n// ─── AIO MARKETING KPIs (the \"math problem\" surface) ────────────────────────\n// CAC = est_marketing_spend / new_customers_acquired\n// LTV = avg_deal_size × retention_months (default retention_months = 12 — configurable)\n// LTV/CAC ratio: healthy SaaS benchmark is ≥ 3:1\n// Funnel conversion rates: TOFU→MOFU (sessions/clicks), MOFU→BOFU (conversions/sessions)\n// mer = Marketing Efficiency Ratio (revenue / spend). Renamed from roi_multiplier. Healthy bench ≥ 3.0.\n// pipeline_coverage = pipeline_value / sales_quota. Healthy ≥ 3.0 (3x quota in pipe).\nexport interface AIOKpis {\n cac: number | null; // Customer Acquisition Cost\n ltv: number | null; // Lifetime Value\n ltv_cac_ratio: number | null; // Higher = better unit economics\n mer: number | null; // Marketing Efficiency Ratio = revenue / marketing spend\n roi_multiplier: number | null; // DEPRECATED alias of mer — kept for backwards compat\n pipeline_coverage: number | null; // pipeline_value / sales_quota\n pipeline_value: number;\n sales_quota: number;\n funnel_conv_tofu_mofu: number | null; // sessions / clicks (%)\n funnel_conv_mofu_bofu: number | null; // conversions / sessions (%)\n conversions: number;\n sessions: number;\n clicks: number;\n marketing_spend: number;\n avg_deal_size: number;\n retention_months: number;\n benchmark_status: {\n ltv_cac: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n mer: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n pipeline_coverage: \"healthy\" | \"borderline\" | \"unhealthy\" | \"unknown\";\n };\n generated_at: string;\n}\n\nasync function getAppConfigNumber(key: string): Promise<number> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [key]);\n if (!r.rows?.length) return 0;\n const v = String((r.rows[0] as any[])[0]).replace(/^\"|\"$/g, \"\");\n return parseFloat(v) || 0;\n } catch { return 0; }\n}\n\nexport async function getAIOKpis(): Promise<AIOKpis> {\n const cached = await getCached(\"aio_kpis\");\n if (cached) return cached;\n\n const summary = await getFunnelSummary();\n const all = summary.stages.flatMap((s) => s.metrics);\n const sessions = getMetric(all, \"ga4\", \"sessions\")?.value || 0;\n const conversions = getMetric(all, \"ga4\", \"conversions\")?.value || 0;\n const clicks = getMetric(all, \"gsc\", \"clicks\")?.value || 0;\n\n const avg_deal_size = await getAppConfigNumber(\"avg_deal_size\");\n const marketing_spend = await getAppConfigNumber(\"est_marketing_spend\");\n const retention_months = (await getAppConfigNumber(\"retention_months\")) || 12;\n const pipeline_value = await getAppConfigNumber(\"pipeline_value\");\n const sales_quota = await getAppConfigNumber(\"sales_quota\");\n\n // Real math, with explicit null when inputs missing.\n const cac = (marketing_spend > 0 && conversions > 0) ? marketing_spend / conversions : null;\n const ltv = avg_deal_size > 0 ? avg_deal_size * retention_months : null;\n const ltv_cac_ratio = (ltv && cac && cac > 0) ? ltv / cac : null;\n // Marketing Efficiency Ratio = revenue / spend. Healthy bench ≥ 3.0 (Klaviyo/Recharge label).\n const mer = (marketing_spend > 0 && conversions > 0 && avg_deal_size > 0)\n ? (conversions * avg_deal_size) / marketing_spend : null;\n const roi_multiplier = mer; // deprecated alias\n // Pipeline Coverage Ratio — sales-team forward-looking signal.\n const pipeline_coverage = (sales_quota > 0 && pipeline_value > 0) ? pipeline_value / sales_quota : null;\n // QA fix: was producing >100% (e.g. 902%) when denominator was very small. Guard with\n // a minimum-sample floor and clamp to 0-100. Anything outside that range means our\n // metric units don't align (clicks vs sessions vs visits), not a real conversion rate.\n const MIN_SAMPLE = 50;\n const funnel_conv_tofu_mofu = (clicks >= MIN_SAMPLE)\n ? Math.max(0, Math.min(100, (sessions / clicks) * 100))\n : null;\n const funnel_conv_mofu_bofu = (sessions >= MIN_SAMPLE)\n ? Math.max(0, Math.min(100, (conversions / sessions) * 100))\n : null;\n\n const benchStatus = (v: number | null, healthy: number, borderline: number) =>\n v == null ? \"unknown\" as const :\n v >= healthy ? \"healthy\" as const :\n v >= borderline ? \"borderline\" as const :\n \"unhealthy\" as const;\n const ltv_cac_status = benchStatus(ltv_cac_ratio, 3, 1);\n const mer_status = benchStatus(mer, 3, 1);\n const pipeline_coverage_status = benchStatus(pipeline_coverage, 3, 1.5);\n\n const out: AIOKpis = {\n cac, ltv, ltv_cac_ratio, mer, roi_multiplier,\n pipeline_coverage, pipeline_value, sales_quota,\n funnel_conv_tofu_mofu, funnel_conv_mofu_bofu,\n conversions, sessions, clicks, marketing_spend, avg_deal_size, retention_months,\n benchmark_status: { ltv_cac: ltv_cac_status, mer: mer_status, pipeline_coverage: pipeline_coverage_status },\n generated_at: new Date().toISOString(),\n };\n await setCached(\"aio_kpis\", out, 300);\n return out;\n}\n\n// ─── INTELLIGENCE FEED (LLM-generated narrative cards) ──────────────────────\nexport async function getIntelligenceFeed(): Promise<IntelligenceCard[]> {\n // Cache invalidation — only serve cached feed if the funnel summary it was built\n // from is not newer than the cache. When the user hits ?force=1 on funnel-summary the\n // upstream cache regenerates, and the feed should follow on its next read instead of\n // serving up-to-5-minute-stale narrative cards.\n const cached = await getCachedTimestamp(\"feed\");\n const summaryMeta = await getCachedTimestamp(\"funnel_summary\");\n if (cached && (!summaryMeta || summaryMeta.generated_at <= cached.generated_at)) {\n return cached.payload;\n }\n\n const summary = await getFunnelSummary();\n const correlations = await getCorrelations();\n\n // Deterministic baseline cards from correlations (no LLM call required)\n const baseCards: IntelligenceCard[] = correlations.slice(0, 4).map((c) => ({\n type: c.direction === \"positive\" ? \"win\" : c.direction === \"negative\" ? \"risk\" : \"correlation\",\n priority: c.strength === \"strong\" ? \"critical\" : c.strength === \"moderate\" ? \"high\" : \"medium\",\n title: c.narrative.split(\".\")[0],\n body: `${c.narrative} Lag: ${c.lag_days}d · confidence: ${Math.round(c.confidence * 100)}%`,\n metrics: [\n `${c.source_metric.integration} ${c.source_metric.metric}: ${c.source_metric.value.toFixed(1)}${c.source_metric.unit}`,\n `${c.target_metric.integration} ${c.target_metric.metric}: ${c.target_metric.value.toFixed(1)}${c.target_metric.unit}`,\n ],\n funnel_stages: c.funnel_stage_from === c.funnel_stage_to ? [c.funnel_stage_from] : [c.funnel_stage_from, c.funnel_stage_to],\n agent_name: \"Signal Monitor\",\n timestamp: new Date().toISOString(),\n action_label: \"Investigate →\",\n action_payload: { question: c.action_prompt, context: { signal_id: c.id, narrative: c.narrative } },\n }));\n\n // Add a bottleneck card\n if (summary.bottleneck) {\n baseCards.unshift({\n type: \"anomaly\", priority: \"critical\",\n title: `${summary.bottleneck.stage} stage is the current bottleneck`,\n body: summary.bottleneck.reason,\n metrics: summary.stages\n .find((s) => s.stage === summary.bottleneck!.stage)!\n .metrics.slice(0, 3)\n .map((m) => `${m.integration} ${m.metric}: ${m.value.toFixed(1)}${m.unit}`),\n funnel_stages: [summary.bottleneck.stage],\n agent_name: \"Funnel Sentinel\",\n timestamp: new Date().toISOString(),\n action_label: \"Diagnose →\",\n action_payload: { question: `What's causing the drop in ${summary.bottleneck.stage} health?` },\n });\n }\n\n // Optionally enrich with LLM narrative when an LLM is configured.\n try {\n const cfg = await getLLMConfig();\n if (cfg.api_key && baseCards.length > 0) {\n const ctx = JSON.stringify({\n stages: summary.stages.map((s) => ({ stage: s.stage, health: s.health_score, metrics: s.metrics.length })),\n correlations: correlations.map((c) => ({ from: c.source_metric.metric, to: c.target_metric.metric, dir: c.direction, narrative: c.narrative })),\n });\n const resp = await callLLM(cfg, [{ role: \"user\", content: `Given this funnel state, write ONE concise 2-sentence \"opportunity\" card body (no preamble, no markdown):\\n${ctx}` }], { maxTokens: 200 });\n if (resp.content) {\n baseCards.splice(1, 0, {\n type: \"opportunity\", priority: \"high\",\n title: \"AI Insight: Cross-stage opportunity detected\",\n body: resp.content.trim(),\n metrics: [\"LLM-derived from funnel state\"],\n funnel_stages: [\"TOFU\", \"MOFU\", \"BOFU\"],\n agent_name: \"Nexus Analyst\",\n timestamp: new Date().toISOString(),\n action_label: \"Explore →\",\n action_payload: { question: \"Tell me more about this opportunity and how to act on it.\" },\n });\n }\n }\n } catch {}\n\n // Empty-state placeholder when nothing connected\n if (baseCards.length === 0) {\n baseCards.push({\n type: \"opportunity\", priority: \"medium\",\n title: \"Connect your first integration to unlock intelligence\",\n body: \"Once Ahrefs, GA4, or Search Console is connected, Nexus will start computing causal signals across your funnel.\",\n metrics: [], funnel_stages: [\"TOFU\"], agent_name: \"Onboarding\",\n timestamp: new Date().toISOString(),\n action_label: \"Open Settings →\",\n action_payload: { question: \"Walk me through connecting my first integration.\" },\n });\n }\n\n await setCached(\"feed\", baseCards, 300);\n return baseCards;\n}\n",[],[365,811],"6bb20ca8-5b48-11f1-9649-ee650bb23af1","integrations.ts","317116ec-56ca-11f1-b821-ee650bb23af1",497,["D",1780050202769],27,"// Integration dispatcher. Every per-integration handler lives in its own file under api/integrations/.\n// This file:\n// 1. exposes `getCredentials` to the router for the OAuth callback\n// 2. dispatches /api/integrations/:name/data to the appropriate per-integration handler\n// 3. handles cross-cutting concerns: Google OAuth fallback, auto-refresh on 401, credential guards\nimport { json, safeJson, getCredentials, withGoogleFallback, withMicrosoftFallback, refreshGoogleToken, refreshMicrosoftToken, GOOGLE_SERVICES, MICROSOFT_SERVICES, MICROSOFT_SCOPES, googleClientId, googleClientSecret, googleRedirectUri, IntegrationCredentials } from \"./integrations/_shared.ts\";\n\nimport * as ahrefs from \"./integrations/ahrefs.ts\";\nimport * as ga4 from \"./integrations/ga4.ts\";\nimport * as gsc from \"./integrations/gsc.ts\";\nimport * as youtube from \"./integrations/youtube.ts\";\nimport * as drive from \"./integrations/drive.ts\";\nimport * as sheets from \"./integrations/sheets.ts\";\nimport * as docs from \"./integrations/docs.ts\";\nimport * as calendar from \"./integrations/calendar.ts\";\nimport * as gmail from \"./integrations/gmail.ts\";\nimport * as linkedin from \"./integrations/linkedin.ts\";\nimport * as outlook from \"./integrations/outlook.ts\";\nimport * as wordpress from \"./integrations/wordpress.ts\";\nimport * as producthunt from \"./integrations/producthunt.ts\";\nimport * as onedrive from \"./integrations/onedrive.ts\";\nimport * as teams from \"./integrations/teams.ts\";\nimport * as excel from \"./integrations/excel.ts\";\n\n// Re-export getCredentials so router.ts (OAuth callback) keeps working without changes.\n// GOOGLE_SERVICES is re-exported so the router can resolve shared Google creds before dispatch.\n// google* helpers are re-exported so the router's OAuth handlers read env-first app credentials.\nexport { getCredentials, GOOGLE_SERVICES, MICROSOFT_SCOPES, googleClientId, googleClientSecret, googleRedirectUri };\nexport type { IntegrationCredentials };\n\n// Map of integration name → handler module. Adding a new integration = adding one file + one entry here.\nconst HANDLERS = {\n ahrefs, ga4, gsc, youtube, drive, sheets, docs, calendar, gmail, linkedin, outlook, wordpress, producthunt,\n onedrive, teams, excel,\n} as const;\n\ntype IntegrationName = keyof typeof HANDLERS;\n\n// Built-in test-connection for integrations whose module didn't export their own `test`.\nasync function defaultTestConnection(name: string, creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n // Google OAuth tokeninfo check + per-service scope coverage report.\n if (name === \"google\" || GOOGLE_SERVICES.includes(name)) {\n const res = await fetch(`https://www.googleapis.com/oauth2/v1/tokeninfo?access_token=${encodeURIComponent(creds.access_token ?? \"\")}`);\n if (!res.ok) return { ok: false, message: \"Invalid or expired access token — click Re-authorize Google\" };\n const info = await safeJson(res);\n const scope = (info.scope ?? \"\") as string;\n // Per-service required scopes (any match = granted). Write-capable scopes checked specifically.\n const SERVICES: Array<[string, string[]]> = [\n [\"GA4\", [\"analytics\"]],\n [\"GSC\", [\"webmasters\", \"search-console\"]],\n [\"YouTube\", [\"youtube\"]],\n [\"Drive RW\", [\"auth/drive \"]],\n [\"Sheets RW\", [\"auth/spreadsheets \"]],\n [\"Docs RW\", [\"auth/documents \"]],\n [\"Calendar\", [\"auth/calendar \"]],\n [\"Gmail RD\", [\"gmail.readonly\", \"gmail.modify\"]],\n [\"Gmail SEND\", [\"gmail.send\"]],\n ];\n const scopeWithTrailingSpace = scope + \" \";\n const granted = SERVICES.filter(([, ks]) => ks.some(k => scopeWithTrailingSpace.includes(k))).map(([n]) => n);\n const missing = SERVICES.filter(([, ks]) => !ks.some(k => scopeWithTrailingSpace.includes(k))).map(([n]) => n);\n const msg = missing.length === 0\n ? `Google verified · all ${SERVICES.length} services + write authorized`\n : `Google verified · ${granted.length}/${SERVICES.length}: ${granted.join(\", \")} · MISSING: ${missing.join(\", \")}`;\n return { ok: missing.length === 0, message: msg };\n }\n return { ok: false, message: \"No test implemented for this integration\" };\n}\n\nexport async function handleIntegrationProxy(\n req: Request,\n name: string,\n searchParams: URLSearchParams,\n creds: IntegrationCredentials\n): Promise<Response> {\n const endpoint = searchParams.get(\"endpoint\") ?? \"\";\n\n // Google services share central OAuth — merge in shared token before dispatching.\n if (GOOGLE_SERVICES.includes(name)) creds = await withGoogleFallback(creds);\n // Microsoft Graph services (onedrive/teams/excel) share the central Outlook token.\n if (MICROSOFT_SERVICES.includes(name)) creds = await withMicrosoftFallback(creds);\n\n // Credential guards\n if ([\"ahrefs\", \"producthunt\"].includes(name) && !creds.api_key && !creds.access_token) {\n return json({ error: `${name} credentials not configured. Add them in Settings.` }, 401);\n }\n if (name === \"youtube\" && !creds.api_key && !creds.access_token) {\n return json({ error: \"YouTube requires an API Key or Google OAuth access token. Configure Google in Settings.\" }, 401);\n }\n if ([\"ga4\", \"gsc\", \"drive\", \"sheets\", \"calendar\", \"gmail\", \"docs\"].includes(name) && !creds.access_token) {\n return json({ error: `${name} requires a Google OAuth access token. Connect with Google in Settings.` }, 401);\n }\n if ([\"linkedin\", \"outlook\", \"onedrive\", \"teams\", \"excel\"].includes(name) && !creds.access_token) {\n const hint = MICROSOFT_SERVICES.includes(name) ? \"Connect Outlook (Microsoft) in Settings.\" : \"Add it in Settings.\";\n return json({ error: `${name} access token not configured. ${hint}` }, 401);\n }\n if (name === \"wordpress\" && !creds.api_key) {\n return json({ error: \"WordPress username not configured. Add credentials in Settings.\" }, 401);\n }\n\n // Test endpoint: prefer integration's own `test` export, fall back to default.\n if (endpoint === \"test\") {\n const handler = HANDLERS[name as IntegrationName];\n const tester = (handler as any)?.test;\n const result = tester ? await tester(creds) : await defaultTestConnection(name, creds);\n return json(result, result.ok ? 200 : 401);\n }\n\n // Google = the credential record itself, no API surface of its own beyond the test endpoint.\n if (name === \"google\") return json({ ok: true, message: \"Google OAuth configured\" });\n\n // Build forwarded query string — exclude our own meta params.\n const excluded = new Set([\"endpoint\", \"integration\"]);\n const forwardedParts: string[] = [];\n searchParams.forEach((value, key) => {\n if (!excluded.has(key)) forwardedParts.push(`${encodeURIComponent(key)}=${encodeURIComponent(value)}`);\n });\n const params = forwardedParts.join(\"&\");\n\n const handler = HANDLERS[name as IntegrationName];\n if (!handler) return json({ error: `Unknown integration: ${name}` }, 400);\n\n // Product Hunt's handler signature takes the URLSearchParams directly (uses GraphQL query param).\n const dispatch = async (c: IntegrationCredentials): Promise<Response> => {\n if (name === \"producthunt\") return (handler as any).handle(req, endpoint, searchParams, c);\n return (handler as any).handle(req, endpoint, params, c);\n };\n\n let resp = await dispatch(creds);\n\n // Auto-refresh Google OAuth on 401, retry once.\n if (resp.status === 401 && GOOGLE_SERVICES.includes(name) && creds.refresh_token) {\n const newToken = await refreshGoogleToken();\n if (newToken) {\n creds = { ...creds, access_token: newToken };\n resp = await dispatch(creds);\n }\n }\n\n // Auto-refresh Microsoft Graph OAuth (outlook + onedrive/teams/excel) on 401, retry once.\n if (resp.status === 401 && (name === \"outlook\" || MICROSOFT_SERVICES.includes(name)) && creds.refresh_token) {\n const newToken = await refreshMicrosoftToken();\n if (newToken) {\n creds = { ...creds, access_token: newToken };\n resp = await dispatch(creds);\n }\n }\n\n return resp;\n}\n",[],[365,820],"66e8d938-5916-11f1-aea8-ee650bb23af1","graphify.ts","66e8da1e-5916-11f1-aea8-ee650bb23af1",288,["D",1779808817573],"export async function getGraphify(): Promise<Record<string, unknown>> {\n return {\n version: \"1.0\",\n app: \"Nexus AI Command Center\",\n graph: {\n pages: [\n { id: \"dashboard\", label: \"Command Center\", component: \"CommandCenter\", routes: [\"/api/intelligence/feed\", \"/api/nsm\", \"/api/aio/cohort-summary\"] },\n { id: \"marketing\", label: \"Strategy\", component: \"MarketingWorkspace\", routes: [\"/api/integrations/ahrefs/data\", \"/api/integrations/ga4/data\", \"/api/integrations/gsc/data\"] },\n { id: \"sales\", label: \"Signals\", component: \"SalesWorkspace\", routes: [\"/api/signals\", \"/api/signals/hot-leads\"] },\n { id: \"workspace\", label: \"Team Hub\", component: \"WorkspacePage\", routes: [] },\n { id: \"objectives\", label: \"OKRs\", component: \"ObjectivesPage\", routes: [\"/api/okrs\", \"/api/okrs/dashboard\"] },\n { id: \"calendar\", label: \"Command Calendar\", component: \"CalendarPage\", routes: [\"/api/calendar/unified\"] },\n { id: \"experiments\", label: \"A/B Tests\", component: \"ExperimentBoard\", routes: [\"/api/experiments\"] },\n { id: \"analytics\", label: \"Analytics\", component: \"AnalyticsPage\", routes: [] },\n { id: \"agents\", label: \"Agent Studio\", component: \"AgentStudio\", routes: [\"/api/agents\"] },\n { id: \"settings\", label: \"Settings\", component: \"SettingsPage\", routes: [\"/api/llm/config\", \"/api/integrations\"] },\n ],\n routes: [\n { method: \"GET\", path: \"/api/health\" },\n { method: \"GET\", path: \"/api/config\" },\n { method: \"GET\", path: \"/api/graphify\" },\n { method: \"GET|POST\", path: \"/api/chat\" },\n { method: \"GET\", path: \"/api/llm/config\" },\n { method: \"POST\", path: \"/api/llm/config\" },\n { method: \"GET\", path: \"/api/agents\" },\n { method: \"POST\", path: \"/api/agents\" },\n { method: \"PUT\", path: \"/api/agents/:id\" },\n { method: \"DELETE\", path: \"/api/agents/:id\" },\n { method: \"GET\", path: \"/api/agents/:id/runs\" },\n { method: \"POST\", path: \"/api/agents/:id/runs\" },\n { method: \"GET\", path: \"/api/integrations\" },\n { method: \"POST\", path: \"/api/integrations/:name\" },\n { method: \"DELETE\", path: \"/api/integrations/:name\" },\n { method: \"GET|POST\", path: \"/api/integrations/:name/data\" },\n { method: \"GET\", path: \"/api/intelligence/funnel-summary\" },\n { method: \"GET\", path: \"/api/intelligence/correlations\" },\n { method: \"GET\", path: \"/api/intelligence/recommendations\" },\n { method: \"GET\", path: \"/api/intelligence/leads\" },\n { method: \"GET\", path: \"/api/intelligence/kpis\" },\n { method: \"GET\", path: \"/api/intelligence/feed\" },\n { method: \"GET|POST\", path: \"/api/signals\" },\n { method: \"PUT\", path: \"/api/signals/:id/action\" },\n { method: \"GET\", path: \"/api/signals/hot-leads\" },\n { method: \"GET\", path: \"/api/signals/hot-leads/count\" },\n { method: \"GET|POST\", path: \"/api/okrs\" },\n { method: \"GET\", path: \"/api/okrs/dashboard\" },\n { method: \"PUT|DELETE\", path: \"/api/okrs/:id\" },\n { method: \"POST\", path: \"/api/okrs/:id/key-results\" },\n { method: \"PUT\", path: \"/api/okrs/key-results/:id\" },\n { method: \"GET|POST\", path: \"/api/nsm\" },\n { method: \"POST\", path: \"/api/nsm/record\" },\n { method: \"GET\", path: \"/api/nsm/progress\" },\n { method: \"GET|POST\", path: \"/api/aio/cohorts\" },\n { method: \"DELETE\", path: \"/api/aio/cohorts/:id\" },\n { method: \"GET\", path: \"/api/aio/cohort-summary\" },\n { method: \"GET\", path: \"/api/aio/engagement-velocity\" },\n { method: \"GET\", path: \"/api/digest\" },\n { method: \"POST\", path: \"/api/digest/generate\" },\n { method: \"GET\", path: \"/api/digest/history\" },\n { method: \"GET|POST\", path: \"/api/experiments\" },\n { method: \"GET\", path: \"/api/calendar/unified\" },\n { method: \"GET\", path: \"/api/tasks\" },\n ],\n integrations: [\n \"google\", \"ahrefs\", \"ga4\", \"gsc\", \"gmail\", \"drive\", \"sheets\",\n \"docs\", \"calendar\", \"linkedin\", \"outlook\", \"wordpress\", \"youtube\", \"producthunt\",\n ],\n agents_schema: {\n fields: [\"id\", \"name\", \"workspace\", \"description\", \"system_prompt\", \"tools\", \"model\", \"enabled\", \"role\"],\n workspaces: [\"marketing\", \"sales\"],\n tool_ids: [\"ahrefs\", \"gsc\", \"ga4\", \"wordpress\", \"youtube\", \"linkedin\", \"outlook\", \"gmail\", \"drive\", \"sheets\", \"docs\", \"producthunt\"],\n },\n features: [\n \"AI chat with tool-calling across integrations\",\n \"Intelligence funnel (TOFU/MOFU/BOFU health scoring)\",\n \"North Star Metric tracking with history\",\n \"OKR management with key results\",\n \"ICE-prioritized A/B experiment board\",\n \"Intent signal detection and hot-lead ranking\",\n \"Cohort LTV analysis\",\n \"Weekly growth digest generation\",\n \"Unified calendar (tasks + experiments + objectives + Google Calendar)\",\n \"Multi-provider LLM config (Anthropic, OpenAI, Gemini, custom)\",\n \"Agent Studio with scheduled runs and budget tracking\",\n \"Command Center terminal with AI-native queries\",\n ],\n },\n };\n}\n",[],[365,829],"201ea062-5d69-11f1-ad33-1607ee4eb77e","experiments.ts","973fc862-589e-11f1-8c3f-ee650bb23af1",611,["D",1780284151688],"// Experiments — Phase 1 Framework Layer.\n// ICE-prioritized experiments mapped to funnel stage + AARRR lever.\n// LLM-assisted ICE scoring uses the funnel summary as live context.\nimport { sqlite } from \"../db.ts\";\nimport { getFunnelSummary } from \"./intelligence.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\nimport { propagateExperimentComplete } from \"./propagate.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type FunnelStage = \"TOFU\" | \"MOFU\" | \"BOFU\";\nexport type AARRRLever =\n | \"acquisition\" | \"activation\" | \"retention\" | \"referral\" | \"revenue\";\n\nexport interface Experiment {\n id: string;\n title: string;\n hypothesis: string | null;\n objective_id: string | null;\n funnel_stage: FunnelStage | null;\n aarrr_lever: AARRRLever | null;\n status: string; // idea | running | completed | killed\n impact: number; // 1..10\n confidence: number; // 1..10\n ease: number; // 1..10\n ice_score: number; // app-computed mean of impact/confidence/ease\n owner: string | null;\n start_date: string | null;\n end_date: string | null;\n channel: string | null;\n success_metric: string | null;\n baseline_value: number | null;\n result_value: number | null;\n outcome: string | null; // win | loss | inconclusive\n learnings: string | null;\n uplift_pct: number | null; // app-computed\n created_at: string;\n updated_at: string;\n}\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return obj;\n}\n\nconst clamp10 = (n: any) => {\n const v = Number(n);\n if (!Number.isFinite(v)) return 5;\n return Math.max(1, Math.min(10, Math.round(v)));\n};\n\nconst computeIce = (impact: number, confidence: number, ease: number) =>\n Math.round(((impact + confidence + ease) / 3) * 100) / 100;\n\nfunction computeUplift(baseline: number | null, result: number | null): number | null {\n if (baseline === null || result === null) return null;\n const b = Number(baseline);\n const r = Number(result);\n if (!Number.isFinite(b) || !Number.isFinite(r) || b === 0) return null;\n return Math.round(((r - b) / Math.abs(b)) * 10000) / 100; // 2 decimal pct\n}\n\nfunction rowToExperiment(columns: string[], row: any[]): Experiment {\n const obj = rowToObj(columns, row);\n const impact = clamp10(obj.impact);\n const confidence = clamp10(obj.confidence);\n const ease = clamp10(obj.ease);\n const baseline = obj.baseline_value != null ? Number(obj.baseline_value) : null;\n const result = obj.result_value != null ? Number(obj.result_value) : null;\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n hypothesis: obj.hypothesis ?? null,\n objective_id: obj.objective_id ?? null,\n funnel_stage: (obj.funnel_stage ?? null) as FunnelStage | null,\n aarrr_lever: (obj.aarrr_lever ?? null) as AARRRLever | null,\n status: String(obj.status ?? \"idea\"),\n impact,\n confidence,\n ease,\n ice_score: computeIce(impact, confidence, ease),\n owner: obj.owner ?? null,\n start_date: obj.start_date ?? null,\n end_date: obj.end_date ?? null,\n channel: obj.channel ?? null,\n success_metric: obj.success_metric ?? null,\n baseline_value: baseline,\n result_value: result,\n outcome: obj.outcome ?? null,\n learnings: obj.learnings ?? null,\n uplift_pct: computeUplift(baseline, result),\n created_at: String(obj.created_at ?? \"\"),\n updated_at: String(obj.updated_at ?? \"\"),\n };\n}\n\n// ─── Read ────────────────────────────────────────────────────────────────────\n\nexport async function listExperiments(filters: {\n status?: string;\n funnel_stage?: string;\n aarrr_lever?: string;\n} = {}): Promise<Experiment[]> {\n try {\n const where: string[] = [];\n const args: any[] = [];\n if (filters.status) { where.push(\"status = ?\"); args.push(filters.status); }\n if (filters.funnel_stage) { where.push(\"funnel_stage = ?\"); args.push(filters.funnel_stage); }\n if (filters.aarrr_lever) { where.push(\"aarrr_lever = ?\"); args.push(filters.aarrr_lever); }\n const sql = `SELECT id, title, hypothesis, objective_id, funnel_stage, aarrr_lever,\n status, impact, confidence, ease, owner, start_date, end_date,\n channel, success_metric, baseline_value, result_value, outcome,\n learnings, created_at, updated_at\n FROM experiments\n ${where.length ? \"WHERE \" + where.join(\" AND \") : \"\"}`;\n const r = await sqlite.execute(sql, args);\n const experiments = (r.rows || []).map(\n (row) => rowToExperiment(r.columns, row as any[]),\n );\n // Sort by ICE DESC in app code.\n experiments.sort((a, b) => b.ice_score - a.ice_score);\n return experiments;\n } catch {\n return [];\n }\n}\n\nasync function getExperimentById(id: string): Promise<Experiment | null> {\n const r = await sqlite.execute(\n `SELECT id, title, hypothesis, objective_id, funnel_stage, aarrr_lever,\n status, impact, confidence, ease, owner, start_date, end_date,\n channel, success_metric, baseline_value, result_value, outcome,\n learnings, created_at, updated_at\n FROM experiments WHERE id = ?`,\n [id],\n );\n if (!r.rows?.length) return null;\n return rowToExperiment(r.columns, r.rows[0] as any[]);\n}\n\n// ─── Write ───────────────────────────────────────────────────────────────────\n\nexport async function createExperiment(\n data: Partial<Experiment>,\n): Promise<Experiment> {\n const now = new Date().toISOString();\n const impact = clamp10(data.impact ?? 5);\n const confidence = clamp10(data.confidence ?? 5);\n const ease = clamp10(data.ease ?? 5);\n\n await sqlite.execute(\n `INSERT INTO experiments\n (title, hypothesis, objective_id, funnel_stage, aarrr_lever, status,\n impact, confidence, ease, owner, start_date, end_date, channel,\n success_metric, baseline_value, result_value, outcome, learnings,\n created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled Experiment\",\n data.hypothesis ?? null,\n data.objective_id ?? null,\n data.funnel_stage ?? null,\n data.aarrr_lever ?? null,\n data.status ?? \"idea\",\n impact, confidence, ease,\n data.owner ?? null,\n data.start_date ?? null,\n data.end_date ?? null,\n data.channel ?? null,\n data.success_metric ?? null,\n data.baseline_value ?? null,\n data.result_value ?? null,\n data.outcome ?? null,\n data.learnings ?? null,\n now, now,\n ],\n );\n const rowIdResult = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((rowIdResult.rows[0] as any[])[0]);\n return (await getExperimentById(id))!;\n}\n\nexport async function updateExperiment(\n id: string,\n data: Partial<Experiment>,\n): Promise<Experiment | null> {\n try {\n const cur = await getExperimentById(id);\n if (!cur) return null;\n const now = new Date().toISOString();\n const next = {\n title: data.title ?? cur.title,\n hypothesis: \"hypothesis\" in data ? (data.hypothesis ?? null) : cur.hypothesis,\n objective_id: \"objective_id\" in data ? (data.objective_id ?? null) : cur.objective_id,\n funnel_stage: \"funnel_stage\" in data ? (data.funnel_stage ?? null) : cur.funnel_stage,\n aarrr_lever: \"aarrr_lever\" in data ? (data.aarrr_lever ?? null) : cur.aarrr_lever,\n status: data.status ?? cur.status,\n impact: data.impact != null ? clamp10(data.impact) : cur.impact,\n confidence: data.confidence != null ? clamp10(data.confidence) : cur.confidence,\n ease: data.ease != null ? clamp10(data.ease) : cur.ease,\n owner: \"owner\" in data ? (data.owner ?? null) : cur.owner,\n start_date: \"start_date\" in data ? (data.start_date ?? null) : cur.start_date,\n end_date: \"end_date\" in data ? (data.end_date ?? null) : cur.end_date,\n channel: \"channel\" in data ? (data.channel ?? null) : cur.channel,\n success_metric: \"success_metric\" in data ? (data.success_metric ?? null) : cur.success_metric,\n baseline_value: \"baseline_value\" in data ? (data.baseline_value ?? null) : cur.baseline_value,\n result_value: \"result_value\" in data ? (data.result_value ?? null) : cur.result_value,\n outcome: \"outcome\" in data ? (data.outcome ?? null) : cur.outcome,\n learnings: \"learnings\" in data ? (data.learnings ?? null) : cur.learnings,\n };\n await sqlite.execute(\n `UPDATE experiments SET\n title = ?, hypothesis = ?, objective_id = ?, funnel_stage = ?, aarrr_lever = ?,\n status = ?, impact = ?, confidence = ?, ease = ?, owner = ?, start_date = ?,\n end_date = ?, channel = ?, success_metric = ?, baseline_value = ?,\n result_value = ?, outcome = ?, learnings = ?, updated_at = ?\n WHERE id = ?`,\n [\n next.title, next.hypothesis, next.objective_id, next.funnel_stage, next.aarrr_lever,\n next.status, next.impact, next.confidence, next.ease, next.owner, next.start_date,\n next.end_date, next.channel, next.success_metric, next.baseline_value,\n next.result_value, next.outcome, next.learnings, now, id,\n ],\n );\n return await getExperimentById(id);\n } catch {\n return null;\n }\n}\n\nexport async function deleteExperiment(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM experiments WHERE id = ?\", [id]);\n return true;\n } catch {\n return false;\n }\n}\n\nexport async function runExperiment(\n id: string,\n baseline_value: number,\n): Promise<Experiment | null> {\n try {\n const now = new Date().toISOString();\n await sqlite.execute(\n `UPDATE experiments\n SET status = 'running', start_date = ?, baseline_value = ?, updated_at = ?\n WHERE id = ?`,\n [now, Number(baseline_value) || 0, now, id],\n );\n return await getExperimentById(id);\n } catch {\n return null;\n }\n}\n\nexport async function completeExperiment(\n id: string,\n data: { result_value: number; outcome: string; learnings?: string },\n): Promise<Experiment | null> {\n try {\n const now = new Date().toISOString();\n await sqlite.execute(\n `UPDATE experiments\n SET status = 'completed', end_date = ?, result_value = ?, outcome = ?,\n learnings = ?, updated_at = ?\n WHERE id = ?`,\n [\n now,\n Number(data.result_value) || 0,\n data.outcome,\n data.learnings ?? null,\n now,\n id,\n ],\n );\n const updated = await getExperimentById(id);\n // Closed-loop: a completed experiment propagates to the measure layer\n // (executed-work counter + calendar marker + linked-objective progress refresh).\n if (updated) {\n await propagateExperimentComplete({\n title: updated.title,\n outcome: updated.outcome,\n objective_id: updated.objective_id,\n });\n }\n return updated;\n } catch {\n return null;\n }\n}\n\n// ─── Stats ───────────────────────────────────────────────────────────────────\n\nexport async function getExperimentStats(): Promise<{\n win_rate: number;\n avg_uplift: number;\n experiments_by_stage: Record<FunnelStage, number>;\n top_lever: AARRRLever | null;\n fastest_days: number | null;\n total: number;\n by_status: Record<string, number>;\n}> {\n const fallback = {\n win_rate: 0,\n avg_uplift: 0,\n experiments_by_stage: { TOFU: 0, MOFU: 0, BOFU: 0 } as Record<FunnelStage, number>,\n top_lever: null as AARRRLever | null,\n fastest_days: null as number | null,\n total: 0,\n by_status: {} as Record<string, number>,\n };\n try {\n const all = await listExperiments();\n if (!all.length) return fallback;\n\n const completed = all.filter((e) => e.status === \"completed\");\n const wins = completed.filter((e) => e.outcome === \"win\").length;\n const win_rate = completed.length > 0\n ? Math.round((wins / completed.length) * 100)\n : 0;\n\n const uplifts = completed\n .map((e) => e.uplift_pct)\n .filter((u): u is number => u !== null && Number.isFinite(u));\n const avg_uplift = uplifts.length > 0\n ? Math.round((uplifts.reduce((a, b) => a + b, 0) / uplifts.length) * 100) / 100\n : 0;\n\n const experiments_by_stage: Record<FunnelStage, number> = { TOFU: 0, MOFU: 0, BOFU: 0 };\n for (const e of all) {\n if (e.funnel_stage && experiments_by_stage[e.funnel_stage] != null) {\n experiments_by_stage[e.funnel_stage] += 1;\n }\n }\n\n // Top lever = lever with highest win count among completed experiments.\n const leverWins = new Map<AARRRLever, number>();\n for (const e of completed) {\n if (e.outcome !== \"win\" || !e.aarrr_lever) continue;\n leverWins.set(e.aarrr_lever, (leverWins.get(e.aarrr_lever) || 0) + 1);\n }\n let top_lever: AARRRLever | null = null;\n let topCount = 0;\n for (const [lever, count] of leverWins) {\n if (count > topCount) { top_lever = lever; topCount = count; }\n }\n\n // Fastest completed experiment (start → end in days).\n let fastest_days: number | null = null;\n for (const e of completed) {\n if (!e.start_date || !e.end_date) continue;\n const days = (new Date(e.end_date).getTime() - new Date(e.start_date).getTime()) / 86400000;\n if (days < 0) continue;\n if (fastest_days === null || days < fastest_days) fastest_days = days;\n }\n if (fastest_days !== null) fastest_days = Math.round(fastest_days * 10) / 10;\n\n const by_status: Record<string, number> = {};\n for (const e of all) {\n by_status[e.status] = (by_status[e.status] || 0) + 1;\n }\n\n return {\n win_rate,\n avg_uplift,\n experiments_by_stage,\n top_lever,\n fastest_days,\n total: all.length,\n by_status,\n };\n } catch {\n return fallback;\n }\n}\n\n// ─── LLM-Assisted ICE Scoring ────────────────────────────────────────────────\n\nconst FALLBACK_ICE = {\n impact: 5,\n confidence: 5,\n ease: 5,\n rationale: \"LLM unavailable — defaulted to neutral ICE.\",\n};\n\nfunction parseICEJson(raw: string): {\n impact: number; confidence: number; ease: number; rationale: string;\n} {\n // First try direct JSON parse, then a regex-fenced fallback (handles\n // ```json ... ``` wrapping or extra prose around the object).\n try {\n const parsed = JSON.parse(raw);\n return {\n impact: clamp10(parsed.impact),\n confidence: clamp10(parsed.confidence),\n ease: clamp10(parsed.ease),\n rationale: String(parsed.rationale ?? \"\").slice(0, 500),\n };\n } catch { /* fall through */ }\n const match = raw.match(/\\{[\\s\\S]*\\}/);\n if (match) {\n try {\n const parsed = JSON.parse(match[0]);\n return {\n impact: clamp10(parsed.impact),\n confidence: clamp10(parsed.confidence),\n ease: clamp10(parsed.ease),\n rationale: String(parsed.rationale ?? \"\").slice(0, 500),\n };\n } catch { /* give up */ }\n }\n return { ...FALLBACK_ICE };\n}\n\nexport async function suggestICE(input: {\n title: string;\n hypothesis?: string;\n funnel_stage?: string;\n aarrr_lever?: string;\n objective_id?: string;\n}): Promise<{\n impact: number;\n confidence: number;\n ease: number;\n rationale: string;\n}> {\n try {\n const config = await getLLMConfig();\n if (!config.api_key) return { ...FALLBACK_ICE };\n\n // Pull funnel summary as context so the model has actual numbers, not just\n // theoretical guesses. Trim to keep prompt small.\n let funnelContext = \"\";\n try {\n const summary = await getFunnelSummary();\n const lines: string[] = [];\n lines.push(`Overall funnel health: ${summary.overall_health}/100`);\n if (summary.bottleneck) {\n lines.push(`Bottleneck stage: ${summary.bottleneck.stage} — ${summary.bottleneck.reason}`);\n }\n for (const s of summary.stages) {\n lines.push(`${s.stage} health=${s.health_score} (${s.metrics.length} metrics)`);\n }\n funnelContext = lines.join(\"\\n\");\n } catch { /* funnel unavailable, score on title alone */ }\n\n const system =\n \"You are a growth analyst scoring an experiment idea using the ICE framework. \" +\n \"Return ONLY valid JSON of shape \" +\n '{\"impact\":1-10,\"confidence\":1-10,\"ease\":1-10,\"rationale\":\"<=2 sentences\"}. ' +\n \"No markdown, no commentary outside the JSON.\";\n\n const user =\n `Experiment: ${input.title}\\n` +\n `Hypothesis: ${input.hypothesis ?? \"(none)\"}\\n` +\n `Funnel stage: ${input.funnel_stage ?? \"(unspecified)\"}\\n` +\n `AARRR lever: ${input.aarrr_lever ?? \"(unspecified)\"}\\n\\n` +\n `Current funnel context:\\n${funnelContext || \"(no live funnel data)\"}\\n\\n` +\n `Score this experiment 1-10 on Impact, Confidence, Ease. Return JSON only.`;\n\n const res = await callLLM(\n config,\n [{ role: \"user\", content: user } as any],\n { systemPrompt: system, maxTokens: 800, temperature: 0.3 },\n );\n const parsed = parseICEJson(res.content || \"\");\n return parsed;\n } catch {\n return { ...FALLBACK_ICE };\n }\n}\n",[],[365,837],"ef0da470-5a5f-11f1-a18f-ee650bb23af1","digest.ts","98641374-589e-11f1-bbad-ee650bb23af1",417,["D",1779950350277],"// Weekly growth digest + daily standup generator.\n// digest = LLM-narrated weekly executive brief stored in growth_digests.\n// standup = ephemeral daily delta brief (not persisted).\n//\n// Both fall back gracefully to deterministic-only assembly when LLM times out\n// or is misconfigured, so the UI never blocks on the model.\nimport { sqlite } from \"../db.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nexport function mondayOf(date: Date | string): string {\n const d = typeof date === \"string\" ? new Date(date) : new Date(date.getTime());\n // getDay: 0=Sun..6=Sat. Shift so Monday = 0.\n const dow = (d.getDay() + 6) % 7;\n d.setDate(d.getDate() - dow);\n d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\n// Run a promise with a hard timeout; resolves to null on timeout.\nfunction withTimeout<T>(p: Promise<T>, ms: number): Promise<T | null> {\n return new Promise<T | null>((resolve) => {\n let done = false;\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null); } }, ms);\n p.then(\n (v) => { if (!done) { done = true; clearTimeout(timer); resolve(v); } },\n () => { if (!done) { done = true; clearTimeout(timer); resolve(null); } },\n );\n });\n}\n\n// Safe dynamic-import wrapper. Returns null if the module doesn't exist or\n// throws on load — lets digest still assemble from whatever data IS available.\nasync function safeImport<T = any>(path: string): Promise<T | null> {\n try { return (await import(path)) as T; } catch { return null; }\n}\n\n// Attempt to parse JSON from an LLM reply, tolerating ```json fences and trailing prose.\nfunction tryParseJson(text: string): any | null {\n if (!text) return null;\n let t = text.trim();\n // Strip markdown fences.\n const fence = t.match(/```(?:json)?\\s*([\\s\\S]*?)```/i);\n if (fence) t = fence[1].trim();\n // Find first { ... last } pair.\n const first = t.indexOf(\"{\");\n const last = t.lastIndexOf(\"}\");\n if (first < 0 || last < first) return null;\n const slice = t.slice(first, last + 1);\n try { return JSON.parse(slice); } catch { return null; }\n}\n\n// ─── Deterministic data collection ───────────────────────────────────────────\n\nasync function collectDigestContext() {\n const nsmMod = await safeImport<any>(\"./nsm.ts\");\n const okrsMod = await safeImport<any>(\"./okrs.ts\");\n const expMod = await safeImport<any>(\"./experiments.ts\");\n const intelMod = await safeImport<any>(\"./intelligence.ts\");\n const signalsMod = await safeImport<any>(\"./signals.ts\");\n\n const [nsm, okr, exp, correlations, hot] = await Promise.all([\n nsmMod?.getNSMProgress?.().catch(() => null) ?? Promise.resolve(null),\n okrsMod?.getOKRDashboard?.().catch(() => null) ?? Promise.resolve(null),\n expMod?.getExperimentStats?.().catch(() => null) ?? Promise.resolve(null),\n intelMod?.getCorrelations?.().catch(() => []) ?? Promise.resolve([]),\n signalsMod?.getHotLeads?.().catch(() => []) ?? Promise.resolve([]),\n ]);\n\n return {\n nsm: nsm || null,\n okr: okr || null,\n experiments: exp || null,\n correlations: Array.isArray(correlations) ? correlations.slice(0, 6) : [],\n hot_leads: Array.isArray(hot) ? hot.slice(0, 5) : [],\n };\n}\n\n// Deterministic fallback digest — used when LLM unavailable.\nfunction deterministicDigest(ctx: any, weekOf: string) {\n const nsmValue = ctx.nsm?.current_value ?? null;\n const nsmTarget = ctx.nsm?.target_value ?? null;\n const nsmStatus = (() => {\n if (!ctx.nsm) return \"No North Star metric configured.\";\n if (nsmValue == null || nsmTarget == null) return \"NSM data unavailable this period.\";\n const pct = nsmTarget > 0 ? (nsmValue / nsmTarget) * 100 : 0;\n return `${ctx.nsm.metric_name ?? \"NSM\"} at ${nsmValue} (${pct.toFixed(0)}% of target ${nsmTarget}).`;\n })();\n\n const positives = (ctx.correlations || []).filter((c: any) => c.direction === \"positive\");\n const negatives = (ctx.correlations || []).filter((c: any) => c.direction === \"negative\");\n const topWin = positives[0]?.narrative ?? \"No strong positive correlations detected this week.\";\n const topRisk = negatives[0]?.narrative ?? \"No critical risks flagged.\";\n\n const recommended_experiments: string[] = [];\n if (negatives[0]?.action_prompt) recommended_experiments.push(negatives[0].action_prompt);\n if (positives[0]?.action_prompt) recommended_experiments.push(positives[0].action_prompt);\n if (recommended_experiments.length === 0) {\n recommended_experiments.push(\"Add 1-2 hypothesis-driven experiments to fill the empty backlog.\");\n }\n\n // getOKRDashboard() returns scalar stats ({total,on_track,at_risk,avg_progress,quarter}),\n // not an objectives array — build callouts from those.\n const okr_callouts: string[] = [];\n if (ctx.okr && (ctx.okr.total ?? 0) > 0) {\n okr_callouts.push(`${ctx.okr.total} objective${ctx.okr.total > 1 ? \"s\" : \"\"}${ctx.okr.quarter ? \" (\" + ctx.okr.quarter + \")\" : \"\"}: ${Math.round(ctx.okr.avg_progress ?? 0)}% avg progress.`);\n okr_callouts.push(`${ctx.okr.on_track ?? 0} on track · ${ctx.okr.at_risk ?? 0} at risk.`);\n } else {\n okr_callouts.push(\"No active objectives — set Q-goals to unlock OKR tracking.\");\n }\n\n const next_actions: string[] = [];\n if ((ctx.hot_leads || []).length) {\n next_actions.push(`Follow up with ${ctx.hot_leads.length} hot lead${ctx.hot_leads.length > 1 ? \"s\" : \"\"} (top: ${ctx.hot_leads[0].name || ctx.hot_leads[0].email}).`);\n }\n if (ctx.experiments?.running_count) {\n next_actions.push(`Review ${ctx.experiments.running_count} running experiment${ctx.experiments.running_count > 1 ? \"s\" : \"\"} for early results.`);\n }\n if (next_actions.length === 0) next_actions.push(\"No urgent items — focus on planning next week's bets.\");\n\n const headline = nsmValue != null && nsmTarget != null\n ? `Week of ${weekOf}: NSM at ${nsmValue}/${nsmTarget}, ${ctx.hot_leads?.length || 0} hot leads.`\n : `Week of ${weekOf}: ${ctx.hot_leads?.length || 0} hot leads, ${(ctx.correlations || []).length} signals tracked.`;\n\n return {\n headline,\n nsm_status: nsmStatus,\n top_win: topWin,\n top_risk: topRisk,\n recommended_experiments,\n okr_callouts,\n next_actions,\n };\n}\n\n// ─── Weekly digest ───────────────────────────────────────────────────────────\n\nexport interface GrowthDigest {\n headline: string;\n nsm_status: string;\n top_win: string;\n top_risk: string;\n recommended_experiments: string[];\n okr_callouts: string[];\n next_actions: string[];\n week_of: string;\n generated_at: string;\n}\n\nexport async function generateDigest(): Promise<GrowthDigest> {\n const weekOf = mondayOf(new Date());\n const ctx = await collectDigestContext();\n const fallback = deterministicDigest(ctx, weekOf);\n\n let enriched: any = null;\n try {\n const cfg = await getLLMConfig();\n if (cfg?.api_key) {\n const prompt = [\n \"You are the growth analyst for an early-stage startup. Produce a weekly growth digest as STRICT JSON.\",\n \"Use only the data provided — never invent metrics.\",\n \"Keys (all required): headline (string, <=120 chars), nsm_status (string), top_win (string), top_risk (string),\",\n \"recommended_experiments (array of 2-4 short strings), okr_callouts (array of 1-3 strings), next_actions (array of 2-4 short strings).\",\n \"Respond with ONLY the JSON object, no prose, no markdown fences.\",\n \"\",\n `WEEK_OF: ${weekOf}`,\n `NSM: ${JSON.stringify(ctx.nsm)}`,\n `OKRS: ${JSON.stringify(ctx.okr)}`,\n `EXPERIMENTS: ${JSON.stringify(ctx.experiments)}`,\n `CORRELATIONS: ${JSON.stringify(ctx.correlations)}`,\n `HOT_LEADS: ${JSON.stringify(ctx.hot_leads)}`,\n ].join(\"\\n\");\n\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 1500,\n temperature: 0.4,\n systemPrompt: \"You write tight, executive-level growth briefs. Always respond with valid JSON.\",\n });\n const result = await withTimeout(llmCall, 30000) as { content?: string } | null;\n if (result?.content) {\n enriched = tryParseJson(result.content);\n }\n }\n } catch { /* LLM failed — fallback already prepared */ }\n\n // Merge: enriched fields override fallback ones; missing keys keep fallback.\n const merged = { ...fallback, ...(enriched || {}) };\n\n // Coerce list fields back to arrays in case the LLM returned strings.\n for (const key of [\"recommended_experiments\", \"okr_callouts\", \"next_actions\"] as const) {\n if (!Array.isArray((merged as any)[key])) {\n (merged as any)[key] = (fallback as any)[key];\n }\n }\n\n const generated_at = new Date().toISOString();\n const digest: GrowthDigest = {\n headline: String(merged.headline ?? fallback.headline),\n nsm_status: String(merged.nsm_status ?? fallback.nsm_status),\n top_win: String(merged.top_win ?? fallback.top_win),\n top_risk: String(merged.top_risk ?? fallback.top_risk),\n recommended_experiments: merged.recommended_experiments,\n okr_callouts: merged.okr_callouts,\n next_actions: merged.next_actions,\n week_of: weekOf,\n generated_at,\n };\n\n // Persist (upsert per week_of so re-runs in the same week replace).\n try {\n const nsmValue = ctx.nsm?.current_value ?? null;\n const existing = await sqlite.execute(\"SELECT id FROM growth_digests WHERE week_of = ?\", [weekOf]);\n if (existing.rows?.length) {\n await sqlite.execute(\n \"UPDATE growth_digests SET payload = ?, nsm_value_at_generation = ?, generated_at = ? WHERE week_of = ?\",\n [JSON.stringify(digest), nsmValue, generated_at, weekOf]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO growth_digests (id, week_of, payload, nsm_value_at_generation, generated_at) VALUES (?, ?, ?, ?, ?)\",\n [crypto.randomUUID(), weekOf, JSON.stringify(digest), nsmValue, generated_at]\n );\n }\n } catch { /* non-critical: still return digest */ }\n\n return digest;\n}\n\nexport async function getLatestDigest(): Promise<GrowthDigest | null> {\n try {\n const r = await sqlite.execute(\n \"SELECT payload, week_of, generated_at FROM growth_digests ORDER BY generated_at DESC LIMIT 1\"\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any[];\n let parsed: any = null;\n try { parsed = JSON.parse(String(row[0])); } catch { parsed = null; }\n if (!parsed) return null;\n return {\n headline: parsed.headline ?? \"\",\n nsm_status: parsed.nsm_status ?? \"\",\n top_win: parsed.top_win ?? \"\",\n top_risk: parsed.top_risk ?? \"\",\n recommended_experiments: Array.isArray(parsed.recommended_experiments) ? parsed.recommended_experiments : [],\n okr_callouts: Array.isArray(parsed.okr_callouts) ? parsed.okr_callouts : [],\n next_actions: Array.isArray(parsed.next_actions) ? parsed.next_actions : [],\n week_of: String(row[1]),\n generated_at: String(row[2]),\n };\n } catch { return null; }\n}\n\nexport async function getDigestHistory(limit = 10): Promise<Array<{ week_of: string; headline: string; generated_at: string }>> {\n try {\n const r = await sqlite.execute(\n \"SELECT week_of, payload, generated_at FROM growth_digests ORDER BY week_of DESC LIMIT ?\",\n [limit]\n );\n return (r.rows || []).map((row: any) => {\n let headline = \"\";\n try { headline = String(JSON.parse(String(row[1]))?.headline ?? \"\"); } catch {}\n return { week_of: String(row[0]), headline, generated_at: String(row[2]) };\n });\n } catch { return []; }\n}\n\n// ─── Daily standup (ephemeral) ───────────────────────────────────────────────\n\nexport interface StandupBrief {\n yesterday: string[];\n today: string[];\n blockers: string[];\n signal_alert: string;\n}\n\nasync function collectStandupContext(workspace: string) {\n // tasks moved to done in last 24h\n let doneTasks: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, completed_at FROM tasks\n WHERE workspace = ? AND status = 'done' AND completed_at >= datetime('now','-1 day')\n ORDER BY completed_at DESC LIMIT 20`,\n [workspace]\n );\n doneTasks = (r.rows || []).map((row: any) => ({\n id: row[0], title: row[1], completed_at: row[2],\n }));\n } catch { doneTasks = []; }\n\n // pending tasks for \"today\" plan\n let pendingTasks: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title FROM tasks\n WHERE workspace = ? AND status IN ('todo','doing')\n ORDER BY created_at DESC LIMIT 20`,\n [workspace]\n );\n pendingTasks = (r.rows || []).map((row: any) => ({ id: row[0], title: row[1] }));\n } catch { pendingTasks = []; }\n\n // experiments with status changes in last 24h (table may not exist yet)\n let expChanges: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, status, updated_at FROM experiments\n WHERE updated_at >= datetime('now','-1 day')\n ORDER BY updated_at DESC LIMIT 10`\n );\n expChanges = (r.rows || []).map((row: any) => ({\n id: row[0], name: row[1], status: row[2], updated_at: row[3],\n }));\n } catch { expChanges = []; }\n\n // new intent signals in 24h\n let newSignals: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT lead_email, lead_name, signal_type, signal_strength FROM intent_signals\n WHERE detected_at >= datetime('now','-1 day')\n ORDER BY signal_strength DESC, detected_at DESC LIMIT 10`\n );\n newSignals = (r.rows || []).map((row: any) => ({\n email: row[0], name: row[1], type: row[2], strength: row[3],\n }));\n } catch { newSignals = []; }\n\n // OKR progress changes in 24h (key_results table may not exist yet)\n let krChanges: any[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, title, progress, updated_at FROM key_results\n WHERE updated_at >= datetime('now','-1 day')\n ORDER BY updated_at DESC LIMIT 10`\n );\n krChanges = (r.rows || []).map((row: any) => ({\n id: row[0], title: row[1], progress: row[2], updated_at: row[3],\n }));\n } catch { krChanges = []; }\n\n return { doneTasks, pendingTasks, expChanges, newSignals, krChanges };\n}\n\nfunction deterministicStandup(ctx: any): StandupBrief {\n const yesterday: string[] = [];\n for (const t of ctx.doneTasks.slice(0, 5)) yesterday.push(`Shipped: ${t.title}`);\n for (const e of ctx.expChanges.slice(0, 3)) yesterday.push(`Experiment \"${e.name}\" -> ${e.status}`);\n for (const kr of ctx.krChanges.slice(0, 2)) yesterday.push(`KR \"${kr.title}\" now ${Math.round(kr.progress || 0)}%`);\n if (yesterday.length === 0) yesterday.push(\"No recorded activity in the last 24 hours.\");\n\n const today: string[] = [];\n for (const t of ctx.pendingTasks.slice(0, 5)) today.push(t.title);\n if (today.length === 0) today.push(\"Nothing queued — pick the next experiment off the backlog.\");\n\n const blockers: string[] = [];\n if (ctx.pendingTasks.length > 10) blockers.push(`${ctx.pendingTasks.length} open tasks — backlog growing.`);\n if (ctx.expChanges.some((e: any) => e.status === \"blocked\" || e.status === \"failed\")) {\n blockers.push(\"One or more experiments need triage.\");\n }\n\n const strongSignals = (ctx.newSignals || []).filter((s: any) => (s.strength ?? 0) >= 3);\n const signal_alert = strongSignals.length\n ? `${strongSignals.length} high-intent signal${strongSignals.length > 1 ? \"s\" : \"\"} in last 24h — top: ${strongSignals[0].name || strongSignals[0].email} (${strongSignals[0].type}).`\n : ctx.newSignals.length\n ? `${ctx.newSignals.length} new signal${ctx.newSignals.length > 1 ? \"s\" : \"\"}, none high-intent yet.`\n : \"No new intent signals in the last 24 hours.\";\n\n return { yesterday, today, blockers, signal_alert };\n}\n\nexport async function generateStandup(opts: { workspace: string }): Promise<StandupBrief> {\n const workspace = opts.workspace || \"marketing\";\n const ctx = await collectStandupContext(workspace);\n const fallback = deterministicStandup(ctx);\n\n let enriched: any = null;\n try {\n const cfg = await getLLMConfig();\n if (cfg?.api_key) {\n const prompt = [\n \"You are a standup coach. Produce a daily standup brief as STRICT JSON.\",\n \"Keys (all required): yesterday (array of strings), today (array of strings), blockers (array of strings), signal_alert (string).\",\n \"Be concise — each list item under 80 chars. Respond with ONLY the JSON object.\",\n \"\",\n `WORKSPACE: ${workspace}`,\n `DONE_LAST_24H: ${JSON.stringify(ctx.doneTasks)}`,\n `PENDING: ${JSON.stringify(ctx.pendingTasks)}`,\n `EXPERIMENT_CHANGES: ${JSON.stringify(ctx.expChanges)}`,\n `NEW_SIGNALS: ${JSON.stringify(ctx.newSignals)}`,\n `KR_PROGRESS: ${JSON.stringify(ctx.krChanges)}`,\n ].join(\"\\n\");\n\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 700,\n temperature: 0.4,\n systemPrompt: \"You write punchy daily standups. Always respond with valid JSON.\",\n });\n const result = await withTimeout(llmCall, 10000) as { content?: string } | null;\n if (result?.content) enriched = tryParseJson(result.content);\n }\n } catch { /* fallback */ }\n\n if (!enriched) return fallback;\n\n return {\n yesterday: Array.isArray(enriched.yesterday) ? enriched.yesterday : fallback.yesterday,\n today: Array.isArray(enriched.today) ? enriched.today : fallback.today,\n blockers: Array.isArray(enriched.blockers) ? enriched.blockers : fallback.blockers,\n signal_alert: typeof enriched.signal_alert === \"string\" ? enriched.signal_alert : fallback.signal_alert,\n };\n}\n",[],[365,845],"f380acb0-5b58-11f1-8321-ee650bb23af1","deals.ts","6e518b64-5b48-11f1-bbb6-ee650bb23af1",551,["D",1780057302563],"import { sqlite } from \"../db.ts\";\nimport { getLeadScores } from \"./intelligence.ts\";\nimport { recordConversion } from \"./aio.ts\";\n\n// CRM pipeline deals. Hybrid: rows seed from qualified leads (getLeadScores) via syncDealsFromLeads,\n// and are fully user-editable (create/edit/drag stage). See db.ts `deals` table.\n\nexport const DEAL_STAGES = [\"lead\", \"qualified\", \"proposal\", \"negotiation\", \"won\", \"lost\"] as const;\n\nconst STAGE_DEFAULT_PROB: Record<string, number> = {\n lead: 10, qualified: 30, proposal: 50, negotiation: 70, won: 100, lost: 0,\n};\n\n// Lead.stage (cold|MQL|SQL) → pipeline stage. Cold leads are skipped on sync (too noisy for a pipeline).\nconst LEAD_STAGE_MAP: Record<string, string> = { MQL: \"qualified\", SQL: \"proposal\" };\n\nfunction rowToObj(columns: string[], row: any[]): Record<string, any> {\n const o: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) o[columns[i]] = row[i];\n return o;\n}\n\nexport async function listDeals(stage?: string): Promise<any[]> {\n const q = stage\n ? \"SELECT * FROM deals WHERE stage = ? ORDER BY updated_at DESC\"\n : \"SELECT * FROM deals ORDER BY updated_at DESC\";\n const r = await sqlite.execute(q, stage ? [stage] : []);\n return (r.rows || []).map((row) => rowToObj(r.columns, row as any[]));\n}\n\nexport async function getDeal(id: string): Promise<any | null> {\n const r = await sqlite.execute(\"SELECT * FROM deals WHERE id = ?\", [id]);\n if (!r.rows || r.rows.length === 0) return null;\n const deal = rowToObj(r.columns, r.rows[0] as any[]);\n // Activity = this contact's intent signals (the deal's lead history).\n const email = deal.contact_email || deal.lead_email;\n let signals: any[] = [];\n if (email) {\n const s = await sqlite.execute(\n \"SELECT signal_type, signal_source, signal_strength, detected_at FROM intent_signals WHERE lead_email = ? ORDER BY detected_at DESC LIMIT 50\",\n [email],\n ).catch(() => null);\n signals = (s?.rows || []).map((row: any) => rowToObj(s!.columns, row));\n }\n return { ...deal, signals };\n}\n\nexport async function createDeal(data: Record<string, any>): Promise<any> {\n const id = crypto.randomUUID();\n const stage = DEAL_STAGES.includes(data.stage) ? data.stage : \"lead\";\n const now = new Date().toISOString();\n await sqlite.execute(\n `INSERT INTO deals (id, title, company, contact_name, contact_email, stage, amount, currency, probability, owner, close_date, source, lead_email, notes, created_at, updated_at, last_activity_at)\n VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)`,\n [\n id, data.title ?? null, data.company ?? null, data.contact_name ?? null, data.contact_email ?? null,\n stage, Number(data.amount) || 0, data.currency ?? \"USD\",\n data.probability != null ? Number(data.probability) : (STAGE_DEFAULT_PROB[stage] ?? 0),\n data.owner ?? null, data.close_date ?? null, data.source ?? \"manual\",\n data.lead_email ?? data.contact_email ?? null, data.notes ?? null, now, now, now,\n ],\n );\n return getDeal(id);\n}\n\nexport async function updateDeal(id: string, data: Record<string, any>): Promise<any | null> {\n const existing = await getDeal(id);\n if (!existing) return null;\n const stage = data.stage != null && DEAL_STAGES.includes(data.stage) ? data.stage : existing.stage;\n // A bare stage move (drag) auto-bumps probability to the stage default unless probability was sent.\n const prob = data.probability != null\n ? Number(data.probability)\n : (data.stage != null && data.stage !== existing.stage ? (STAGE_DEFAULT_PROB[stage] ?? existing.probability) : existing.probability);\n const merged = {\n title: data.title ?? existing.title,\n company: data.company ?? existing.company,\n contact_name: data.contact_name ?? existing.contact_name,\n contact_email: data.contact_email ?? existing.contact_email,\n stage,\n amount: data.amount != null ? Number(data.amount) : existing.amount,\n currency: data.currency ?? existing.currency,\n probability: prob,\n owner: data.owner ?? existing.owner,\n close_date: data.close_date ?? existing.close_date,\n notes: data.notes ?? existing.notes,\n };\n await sqlite.execute(\n `UPDATE deals SET title=?, company=?, contact_name=?, contact_email=?, stage=?, amount=?, currency=?, probability=?, owner=?, close_date=?, notes=?, updated_at=?, last_activity_at=? WHERE id=?`,\n [merged.title, merged.company, merged.contact_name, merged.contact_email, merged.stage, merged.amount,\n merged.currency, merged.probability, merged.owner, merged.close_date, merged.notes,\n new Date().toISOString(), new Date().toISOString(), id],\n );\n // Revenue interconnection: when a deal is newly won, record its amount as a conversion so it\n // flows into attribution → the funnel \"attribution/revenue\" metric → the revenue OKR Key Result.\n if (merged.stage === \"won\" && existing.stage !== \"won\" && merged.amount > 0) {\n await recordConversion({\n utm_source: merged.source || \"crm\",\n utm_medium: \"deal\",\n utm_campaign: merged.company || merged.title || undefined,\n conversion_value: merged.amount,\n }).catch(() => {});\n }\n // Symmetry: un-winning a deal backs the revenue out so the OKR stays honest.\n if (existing.stage === \"won\" && merged.stage !== \"won\" && existing.amount > 0) {\n await recordConversion({\n utm_source: existing.source || \"crm\",\n utm_medium: \"deal\",\n utm_campaign: existing.company || existing.title || undefined,\n conversion_value: -existing.amount,\n }).catch(() => {});\n }\n return getDeal(id);\n}\n\nexport async function deleteDeal(id: string): Promise<boolean> {\n const r = await sqlite.execute(\"DELETE FROM deals WHERE id = ?\", [id]);\n return Number((r as any)?.rowsAffected ?? 0) > 0;\n}\n\n// Heuristic: is this email a system/notification sender rather than a real prospect? Used to\n// PRE-deselect junk in the interactive import (the user still decides). Imperfect by design.\nexport function looksLikeSystemSender(email: string): boolean {\n const e = String(email || \"\").toLowerCase();\n const local = e.split(\"@\")[0] || \"\";\n const domain = e.split(\"@\")[1] || \"\";\n const localBad = /(^|[._-])(no-?reply|donotreply|do-not-reply|notif(y|ication)?s?|mailer-daemon|postmaster|support|help(desk)?|system|sysadmin|admin|alerts?|info|hello|team|automated|auto|bot|updates?|news(letter)?|billing|account|security|root|webmaster|invites?|calendar)([._-]|$)/.test(local);\n const domainBad = /(microsoft\\.com|sap\\.com|read\\.ai|google\\.com|atlassian\\.|sendgrid|mailchimp|amazonses|salesforce\\.com|zoom\\.us|calendly|docusign|slack\\.com|github\\.com)/.test(domain)\n || /^(mail|email|notif(y|ications?)?|engage|reply|bounce|news|info)\\./.test(domain);\n return localBad || domainBad;\n}\n\nasync function insertDealFromLead(l: any, have: Set<string>, now: string): Promise<boolean> {\n if (!l.email || have.has(l.email)) return false;\n const stage = LEAD_STAGE_MAP[l.stage] || \"lead\";\n await sqlite.execute(\n `INSERT INTO deals (id, title, company, contact_name, contact_email, stage, amount, probability, source, lead_email, last_activity_at, created_at, updated_at)\n VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)`,\n [crypto.randomUUID(), `${l.company || l.name || l.email} — opportunity`, l.company ?? null, l.name ?? null,\n l.email, stage, 0, STAGE_DEFAULT_PROB[stage] ?? 0, l.source || \"lead\", l.email, l.last_touch_at ?? null, now, now],\n );\n have.add(l.email);\n return true;\n}\n\n// Interactive import: list leads not yet in the pipeline, annotated with a prospect guess so the\n// user can review + select (instead of auto-adding notification/system senders).\nexport async function listCandidates(): Promise<any[]> {\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n return (leads as any[]).filter((l) => l.email && !have.has(l.email)).map((l) => {\n const sys = looksLikeSystemSender(l.email);\n return {\n email: l.email, name: l.name, company: l.company, stage: l.stage,\n mql_score: l.mql_score, sql_signals: l.sql_signals || [], last_touch_at: l.last_touch_at,\n likely_prospect: !sys && l.stage !== \"cold\",\n reason: sys ? \"Looks like a system/notification sender\" : (l.stage === \"cold\" ? \"Low engagement (cold)\" : \"Engaged lead\"),\n };\n });\n}\n\n// Create deals only for the explicitly-selected lead emails (the interactive import).\nexport async function importDeals(emails: string[]): Promise<{ created: number }> {\n const want = new Set((emails || []).map(String));\n if (!want.size) return { created: 0 };\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n const now = new Date().toISOString();\n let created = 0;\n for (const l of leads as any[]) {\n if (!want.has(l.email)) continue;\n if (await insertDealFromLead(l, have, now)) created++;\n }\n return { created };\n}\n\n// Auto-seed deals from qualified (MQL/SQL) leads that don't already have a deal. Idempotent.\n// Kept for convenience / chat; the interactive import (listCandidates + importDeals) is the primary path.\nexport async function syncDealsFromLeads(): Promise<{ created: number; leads: number }> {\n const leads = await getLeadScores().catch(() => []);\n const existing = await sqlite.execute(\"SELECT lead_email FROM deals WHERE lead_email IS NOT NULL\");\n const have = new Set((existing.rows || []).map((r: any[]) => String(r[0])));\n const now = new Date().toISOString();\n let created = 0;\n for (const l of leads as any[]) {\n if (l.stage === \"cold\" || looksLikeSystemSender(l.email)) continue; // skip cold + obvious system senders\n if (await insertDealFromLead(l, have, now)) created++;\n }\n return { created, leads: (leads as any[]).length };\n}\n",[],[365,853],"d435f1b6-5d61-11f1-9b2a-1607ee4eb77e","company.ts","d435f260-5d61-11f1-9b2a-1607ee4eb77e",577,["D",1780281017861],"// Part 15.2 — Company Profile: the app's first-class understanding of the BUSINESS.\n// Until now every prompt described the APP's mechanics, never WHO the company is. This stores\n// a structured profile (positioning, ICP, voice, products, competitors, goals) in app_config\n// under one key, and renders a compact markdown block injected GLOBALLY into every AI surface.\n// It supersedes Part 1 D's flat \"Brand Context\" by giving it structure (with a text alias kept).\nimport { sqlite } from \"../db.ts\";\n\nconst KEY = \"company_profile\";\n\nexport interface CompanyProfile {\n name?: string;\n product?: string; // what it sells (products/services)\n icp?: string; // ideal customer profile, segments, personas\n value_proposition?: string;// positioning / value prop\n competitors?: string;\n voice?: string; // tone/voice + brand do's & don'ts\n differentiators?: string;\n pricing?: string;\n goals?: string; // goals / targets\n markets?: string; // markets / geos\n // free-form fallback (Part 1 D \"Brand Context\" text); used if structured fields are empty.\n notes?: string;\n}\n\nexport async function getCompanyProfile(): Promise<CompanyProfile> {\n try {\n const r = await sqlite.execute(\"SELECT value FROM app_config WHERE key = ?\", [KEY]);\n const raw = (r.rows?.[0] as any[])?.[0];\n if (!raw) return {};\n const parsed = JSON.parse(String(raw));\n return (parsed && typeof parsed === \"object\") ? parsed : {};\n } catch { return {}; }\n}\n\nexport async function setCompanyProfile(data: CompanyProfile): Promise<CompanyProfile> {\n // Merge with existing so partial PUTs don't wipe other fields.\n const cur = await getCompanyProfile();\n const merged: CompanyProfile = { ...cur };\n for (const [k, v] of Object.entries(data || {})) {\n if (typeof v === \"string\") merged[k as keyof CompanyProfile] = v as any;\n }\n await sqlite.execute(\n `INSERT INTO app_config (key, value, updated_at) VALUES (?, ?, datetime('now'))\n ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = datetime('now')`,\n [KEY, JSON.stringify(merged)]\n );\n return merged;\n}\n\n// Part 1 D alias — Brand Context as plain text round-trips into `notes`.\nexport async function getBrandContext(): Promise<string> {\n const p = await getCompanyProfile();\n return (p.notes ?? \"\").toString();\n}\nexport async function setBrandContext(text: string): Promise<void> {\n await setCompanyProfile({ notes: String(text ?? \"\") });\n}\n\nconst LABELS: Array<[keyof CompanyProfile, string]> = [\n [\"name\", \"Company\"],\n [\"product\", \"What we sell\"],\n [\"value_proposition\", \"Positioning / value prop\"],\n [\"icp\", \"Ideal customer (ICP)\"],\n [\"differentiators\", \"Differentiators\"],\n [\"competitors\", \"Competitors\"],\n [\"pricing\", \"Pricing\"],\n [\"markets\", \"Markets\"],\n [\"goals\", \"Goals / targets\"],\n [\"voice\", \"Voice / tone\"],\n];\n\n// Compact markdown for prompt injection (capped ~2KB per Part 15.4 token budget).\nexport async function getCompanyContextString(maxChars = 2000): Promise<string> {\n const p = await getCompanyProfile();\n const lines: string[] = [];\n for (const [k, label] of LABELS) {\n const v = (p[k] ?? \"\").toString().trim();\n if (v) lines.push(`- ${label}: ${v}`);\n }\n if (p.notes && p.notes.trim() && lines.length === 0) {\n // Pure free-form brand context.\n return (\"## Company Profile\\n\" + p.notes.trim()).slice(0, maxChars);\n }\n if (lines.length === 0) return \"\";\n let out = \"## Company Profile (who this business is — make every output specific to it)\\n\" + lines.join(\"\\n\");\n if (p.notes && p.notes.trim()) out += `\\n- Notes: ${p.notes.trim()}`;\n return out.slice(0, maxChars);\n}\n\nexport function isProfileEmpty(p: CompanyProfile): boolean {\n return !Object.values(p || {}).some((v) => typeof v === \"string\" && v.trim());\n}\n",[],[365,861],"1ef55604-5d69-11f1-b78e-1607ee4eb77e","chat.ts","31ca5a90-56ca-11f1-a61d-ee650bb23af1",610,["D",1780284149737],38,"import { sqlite } from \"../db.ts\";\nimport { getLLMConfig, callLLM, callLLMStream } from \"./llm.ts\";\nimport { getFunnelSummary, getCorrelations } from \"./intelligence.ts\";\nimport { listAgents, createAgent, getAgent } from \"./agents.ts\";\nimport { getCredentials, handleIntegrationProxy } from \"./integrations.ts\";\nimport { enqueueAction } from \"./actions.ts\";\nimport { createTask } from \"./tasks.ts\";\nimport { createCalendarEntry } from \"./calendar_unified.ts\";\nimport { listObjectives, createObjective } from \"./okrs.ts\";\nimport { createExperiment } from \"./experiments.ts\";\nimport type { ChatMessage } from \"../types.ts\";\n\n// ─── Constants ────────────────────────────────────────────────────────────────\n\nconst CORS = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, PUT, DELETE, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n};\n\nconst DEFAULT_SYSTEM_PROMPT =\n \"You are an AI assistant for AIHive Command Center, a marketing and sales platform. \" +\n \"You can analyze data, create agents, and execute tasks across integrations.\";\n\n// Grounds the model in its REAL execution capabilities so it actually uses the tools instead of\n// hallucinating that it is \"blocked\" or \"needs credentials\" (the platform already holds them).\nconst EXECUTION_GUIDE =\n \"EXECUTION CAPABILITIES — IMPORTANT: You can actually DO work via tools. NEVER claim you are \" +\n \"'blocked', that you 'need credentials/site details', or that there is a 'security boundary you cannot bypass' — \" +\n \"the platform already stores all integration credentials. \" +\n \"PUBLIC/IRREVERSIBLE actions QUEUE for one-click human approval in the Approvals inbox — call the matching tool: \" +\n \"prepare_wordpress_post (publish to WordPress), prepare_email (send via Gmail), prepare_outlook_email (send via Outlook), prepare_linkedin_post (post to LinkedIn). \" +\n \"INTERNAL/REVERSIBLE actions run IMMEDIATELY (no approval): create_campaign (start a campaign/project), create_task and schedule_calendar_event (create work), \" +\n \"create_drive_folder / create_google_doc / create_google_sheet (create files in Google Drive/Docs/Sheets). \" +\n \"Use get_integration_data to read live data. None of these require the user to provide credentials. \" +\n \"When the user asks you to publish or send something, call the appropriate prepare_* tool with the FULL content, then tell them it is queued in Approvals. \" +\n \"When the user asks to create a campaign/project or a file, call the matching create_* tool and confirm it was created. \" +\n \"Do NOT fabricate step-by-step 'execution logs', and do NOT ask the user for credentials.\";\n\n// ─── Helpers ──────────────────────────────────────────────────────────────────\n\nfunction json(data: any, status = 200): Response {\n return new Response(JSON.stringify(data), {\n status,\n headers: { \"Content-Type\": \"application/json\", ...CORS },\n });\n}\n\nfunction rowToConversation(\n columns: string[],\n row: any[]\n): { id: string; messages: ChatMessage[]; agent_id?: string; workspace?: string } {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n\n let messages: ChatMessage[] = [];\n if (obj.messages) {\n try {\n messages = JSON.parse(obj.messages);\n } catch {\n messages = [];\n }\n }\n\n return {\n id: obj.id,\n messages,\n agent_id: obj.agent_id ?? undefined,\n workspace: obj.workspace ?? undefined,\n };\n}\n\n// ─── Conversation DB ops ──────────────────────────────────────────────────────\n\nasync function loadConversation(\n conversationId: string\n): Promise<{ id: string; messages: ChatMessage[]; agent_id?: string; workspace?: string } | null> {\n const result = await sqlite.execute(\n \"SELECT id, messages, agent_id, workspace FROM conversations WHERE id = ?\",\n [conversationId]\n );\n\n if (!result.rows || result.rows.length === 0) {\n return null;\n }\n\n return rowToConversation(result.columns, result.rows[0]);\n}\n\nasync function saveConversation(\n id: string,\n messages: ChatMessage[],\n agentId?: string,\n workspace?: string\n): Promise<void> {\n const now = new Date().toISOString();\n const messagesJson = JSON.stringify(messages);\n\n const existing = await sqlite.execute(\n \"SELECT id FROM conversations WHERE id = ?\",\n [id]\n );\n\n const ws = workspace ?? \"general\";\n if (existing.rows && existing.rows.length > 0) {\n await sqlite.execute(\n \"UPDATE conversations SET messages = ?, agent_id = ?, workspace = ?, updated_at = ? WHERE id = ?\",\n [messagesJson, agentId ?? null, ws, now, id]\n );\n } else {\n await sqlite.execute(\n `INSERT INTO conversations (id, messages, agent_id, workspace, created_at, updated_at)\n VALUES (?, ?, ?, ?, ?, ?)`,\n [id, messagesJson, agentId ?? null, ws, now, now]\n );\n }\n}\n\n// ─── Tool definitions ─────────────────────────────────────────────────────────\n\nexport function buildTools(): any[] {\n return [\n {\n type: \"function\",\n function: {\n name: \"get_integration_data\",\n description:\n \"Fetch data from ANY connected integration. Marketing/SEO: ahrefs, ga4, gsc, youtube, wordpress, producthunt, linkedin. \" +\n \"Microsoft 365 (all share one connected account): outlook (inbox/calendar/contacts), onedrive (files), \" +\n \"teams (teams/channels/messages/chats), excel (workbooks/worksheets/range). Google Workspace: gmail, drive, sheets, docs, calendar.\",\n parameters: {\n type: \"object\",\n properties: {\n name: {\n type: \"string\",\n description:\n \"Integration name: ahrefs, ga4, gsc, linkedin, outlook, onedrive, teams, excel, wordpress, youtube, producthunt, gmail, drive, sheets, docs, calendar\",\n },\n endpoint: {\n type: \"string\",\n description:\n \"The endpoint within the integration, e.g. onedrive 'recent'/'files'/'search'; teams 'teams'/'channels'/'messages'/'chats'; excel 'workbooks'/'worksheets'/'range'; outlook 'inbox'/'calendar'/'contacts'; ahrefs 'site-explorer'/'keywords'; gsc 'search-analytics'; linkedin 'posts'.\",\n },\n params: {\n type: \"object\",\n description: \"Additional query parameters to forward to the API\",\n additionalProperties: { type: \"string\" },\n },\n },\n required: [\"name\", \"endpoint\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_agent\",\n description:\n \"Create a new AI agent in AIHive Command Center with a specific role, workspace, and toolset.\",\n parameters: {\n type: \"object\",\n properties: {\n name: {\n type: \"string\",\n description: \"Name for the new agent\",\n },\n workspace: {\n type: \"string\",\n enum: [\"marketing\", \"sales\"],\n description: \"The workspace this agent belongs to\",\n },\n description: {\n type: \"string\",\n description: \"Short description of what this agent does\",\n },\n system_prompt: {\n type: \"string\",\n description: \"System prompt that defines the agent's behavior\",\n },\n tools: {\n type: \"array\",\n items: { type: \"string\" },\n description:\n \"List of tool names the agent has access to (e.g. ['get_integration_data', 'compose_content'])\",\n },\n },\n required: [\"name\", \"workspace\", \"description\", \"system_prompt\", \"tools\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"list_agents\",\n description: \"List all AI agents, optionally filtered by workspace.\",\n parameters: {\n type: \"object\",\n properties: {\n workspace: {\n type: \"string\",\n enum: [\"marketing\", \"sales\"],\n description: \"Filter agents by workspace (optional)\",\n },\n },\n required: [],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"list_objectives\",\n description: \"List OKR objectives (id + title). Use before create_campaign to link a campaign to an existing objective.\",\n parameters: {\n type: \"object\",\n properties: {\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"], description: \"Filter by workspace (optional)\" },\n },\n required: [],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"compose_content\",\n description:\n \"Generate marketing or sales content such as blog posts, LinkedIn updates, emails, or YouTube descriptions.\",\n parameters: {\n type: \"object\",\n properties: {\n type: {\n type: \"string\",\n enum: [\"blog\", \"linkedin\", \"email\", \"youtube_description\"],\n description: \"The type of content to generate\",\n },\n topic: {\n type: \"string\",\n description: \"The topic or subject of the content\",\n },\n tone: {\n type: \"string\",\n description:\n \"The tone of voice for the content (e.g. 'professional', 'casual', 'persuasive')\",\n },\n },\n required: [\"type\", \"topic\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_email\",\n description:\n \"Draft an email and QUEUE it for human approval (Gmail). It is NOT sent until the user approves it in Approvals. Use this to actually act on email outreach.\",\n parameters: {\n type: \"object\",\n properties: {\n to: { type: \"string\", description: \"Recipient email address\" },\n subject: { type: \"string\", description: \"Email subject\" },\n body: { type: \"string\", description: \"Email body (plain text)\" },\n cc: { type: \"string\", description: \"Optional CC address\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once sent\" },\n },\n required: [\"to\", \"subject\", \"body\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_linkedin_post\",\n description:\n \"Draft a LinkedIn post and QUEUE it for human approval. NOT posted until the user approves.\",\n parameters: {\n type: \"object\",\n properties: {\n text: { type: \"string\", description: \"The post text\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once posted\" },\n },\n required: [\"text\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_wordpress_post\",\n description:\n \"Draft a WordPress blog post and QUEUE it for human approval. NOT published until the user approves.\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\", description: \"Post title\" },\n content: { type: \"string\", description: \"Post HTML/markdown content\" },\n status: { type: \"string\", enum: [\"publish\", \"draft\"], description: \"publish (default) or draft\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once published\" },\n },\n required: [\"title\", \"content\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_task\",\n description:\n \"Create a task in the task board. Runs immediately (internal, no approval). Optionally link it to an objective, key result, or experiment.\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\" },\n description: { type: \"string\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n priority: { type: \"string\", enum: [\"low\", \"medium\", \"high\", \"urgent\"] },\n due_date: { type: \"string\", description: \"YYYY-MM-DD\" },\n objective_id: { type: \"string\" },\n key_result_id: { type: \"string\" },\n experiment_id: { type: \"string\" },\n },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"schedule_calendar_event\",\n description:\n \"Add an event to the unified calendar. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\" },\n event_date: { type: \"string\", description: \"YYYY-MM-DD\" },\n event_time: { type: \"string\", description: \"HH:MM (optional)\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n },\n required: [\"title\", \"event_date\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_campaign\",\n description:\n \"Create a marketing/sales campaign (a campaign IS an agent). Runs immediately (internal, no approval). \" +\n \"A campaign is interconnected: it links to an OKR objective and gets an ICE experiment so its impact is measured. \" +\n \"Pass objective_id to link an existing objective (use list_objectives first), or omit to auto-create one from objective_title. \" +\n \"Always provide impact/confidence/ease (1-10) so the linked ICE experiment is scored.\",\n parameters: {\n type: \"object\",\n properties: {\n name: { type: \"string\", description: \"Campaign name\" },\n workspace: { type: \"string\", enum: [\"marketing\", \"sales\"] },\n channel: { type: \"string\", description: \"linkedin | email | content | blog | seo | social | paid | mixed\" },\n goal: { type: \"string\", description: \"What the campaign aims to achieve\" },\n audience: { type: \"string\", description: \"Target audience (optional)\" },\n status: { type: \"string\", enum: [\"active\", \"paused\", \"draft\", \"completed\"], description: \"default active\" },\n objective_id: { type: \"string\", description: \"Existing OKR objective id to link (optional)\" },\n objective_title: { type: \"string\", description: \"If no objective_id, auto-create an objective with this title (optional)\" },\n quarter: { type: \"string\", description: \"Quarter for the auto-created objective, e.g. 'Q3 2025' (optional; defaults to current quarter)\" },\n funnel_stage: { type: \"string\", enum: [\"TOFU\", \"MOFU\", \"BOFU\"], description: \"Funnel stage for the ICE experiment\" },\n impact: { type: \"number\", description: \"ICE impact 1-10\" },\n confidence: { type: \"number\", description: \"ICE confidence 1-10\" },\n ease: { type: \"number\", description: \"ICE ease 1-10\" },\n },\n required: [\"name\", \"channel\", \"goal\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_drive_folder\",\n description: \"Create a folder in Google Drive. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: { name: { type: \"string\", description: \"Folder name\" } },\n required: [\"name\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_google_doc\",\n description: \"Create a Google Doc, optionally with initial content. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: {\n title: { type: \"string\", description: \"Document title\" },\n content: { type: \"string\", description: \"Optional initial body text\" },\n },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"create_google_sheet\",\n description: \"Create a Google Sheet. Runs immediately (internal, no approval).\",\n parameters: {\n type: \"object\",\n properties: { title: { type: \"string\", description: \"Spreadsheet title\" } },\n required: [\"title\"],\n },\n },\n },\n {\n type: \"function\",\n function: {\n name: \"prepare_outlook_email\",\n description:\n \"Draft an email and QUEUE it for human approval (Outlook/Microsoft 365). NOT sent until the user approves it in Approvals.\",\n parameters: {\n type: \"object\",\n properties: {\n to: { type: \"string\", description: \"Recipient email address\" },\n subject: { type: \"string\", description: \"Email subject\" },\n body: { type: \"string\", description: \"Email body (plain text)\" },\n cc: { type: \"string\", description: \"Optional CC address\" },\n task_id: { type: \"string\", description: \"Optional task id to mark done once sent\" },\n },\n required: [\"to\", \"subject\", \"body\"],\n },\n },\n },\n ];\n}\n\n// ─── Tool executor ────────────────────────────────────────────────────────────\n\n// Run a WRITE/mutate endpoint on an integration immediately, POSTing through the same proxy\n// the approval queue uses. Returns parsed JSON (or {error}).\nasync function runIntegrationWrite(integration: string, endpoint: string, payload: any): Promise<any> {\n const creds = await getCredentials(integration);\n if (!creds) return { error: `Integration '${integration}' is not connected. Add it in Settings first.` };\n const params = new URLSearchParams({ endpoint });\n const req = new Request(\n `https://aihive.internal/api/integrations/${integration}/data?endpoint=${encodeURIComponent(endpoint)}`,\n { method: \"POST\", headers: { \"Content-Type\": \"application/json\" }, body: JSON.stringify(payload ?? {}) },\n );\n try {\n const res = await handleIntegrationProxy(req, integration, params, creds);\n return await res.json().catch(() => ({}));\n } catch (err: any) {\n return { error: err?.message ?? \"Integration call failed\" };\n }\n}\n\nexport async function executeTool(\n toolName: string,\n toolArgs: Record<string, any>,\n llmConfig: Awaited<ReturnType<typeof getLLMConfig>>\n): Promise<string> {\n switch (toolName) {\n case \"get_integration_data\": {\n const { name, endpoint, params = {} } = toolArgs as {\n name: string;\n endpoint: string;\n params?: Record<string, string>;\n };\n\n const creds = await getCredentials(name);\n if (!creds) {\n return JSON.stringify({\n error: `Integration '${name}' is not configured. Please add credentials first.`,\n });\n }\n\n // Build a synthetic URLSearchParams from args\n const searchParams = new URLSearchParams({ endpoint });\n for (const [k, v] of Object.entries(params)) {\n searchParams.set(k, String(v));\n }\n\n // Build a minimal Request so handlers can read body if needed\n const syntheticReq = new Request(\"https://aihive.internal/proxy\", {\n method: \"GET\",\n });\n\n try {\n const proxyRes = await handleIntegrationProxy(\n syntheticReq,\n name,\n searchParams,\n creds\n );\n const data = await proxyRes.json();\n return JSON.stringify(data);\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Proxy request failed\" });\n }\n }\n\n case \"create_agent\": {\n const { name, workspace, description, system_prompt, tools } = toolArgs as {\n name: string;\n workspace: \"marketing\" | \"sales\";\n description: string;\n system_prompt: string;\n tools: string[];\n };\n\n try {\n const agent = await createAgent({\n name,\n workspace,\n description,\n system_prompt,\n tools,\n });\n return JSON.stringify({ success: true, agent });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create agent\" });\n }\n }\n\n case \"list_agents\": {\n const { workspace } = toolArgs as { workspace?: string };\n try {\n const agents = await listAgents(workspace);\n return JSON.stringify({ agents });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to list agents\" });\n }\n }\n\n case \"compose_content\": {\n const { type, topic, tone = \"professional\" } = toolArgs as {\n type: \"blog\" | \"linkedin\" | \"email\" | \"youtube_description\";\n topic: string;\n tone?: string;\n };\n\n const contentPrompts: Record<string, string> = {\n blog: `Write a comprehensive, SEO-friendly blog post about: \"${topic}\". Tone: ${tone}. Include an introduction, 3-5 sections with headers, and a conclusion.`,\n linkedin: `Write an engaging LinkedIn post about: \"${topic}\". Tone: ${tone}. Keep it under 300 words, include relevant hashtags, and a call-to-action.`,\n email: `Write a professional email about: \"${topic}\". Tone: ${tone}. Include subject line, greeting, body paragraphs, and a clear call-to-action.`,\n youtube_description: `Write an optimized YouTube video description for a video about: \"${topic}\". Tone: ${tone}. Include a hook, timestamps placeholder, links section, and relevant tags.`,\n };\n\n const prompt = contentPrompts[type] ?? `Write content about: \"${topic}\". Tone: ${tone}.`;\n\n try {\n const contentMessages: ChatMessage[] = [\n { role: \"user\", content: prompt },\n ];\n\n const result = await callLLM(llmConfig, contentMessages, {\n systemPrompt:\n \"You are an expert content writer for marketing and sales teams. Produce high-quality, engaging content tailored to the specified format and tone.\",\n maxTokens: 2048,\n });\n\n return JSON.stringify({\n type,\n topic,\n tone,\n content: result.content,\n });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to compose content\" });\n }\n }\n\n case \"prepare_email\": {\n const { to, subject, body, cc, task_id } = toolArgs as any;\n if (!to || !subject) return JSON.stringify({ error: \"to + subject required\" });\n const action = await enqueueAction({\n integration: \"gmail\", endpoint: \"send\",\n payload: { to, subject, body: body ?? \"\", cc },\n title: `Email: ${subject}`,\n preview: `To: ${to}\\nSubject: ${subject}\\n\\n${String(body ?? \"\").slice(0, 600)}`,\n task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"Email drafted and QUEUED FOR APPROVAL — it will NOT send until the user approves it in Approvals.\" });\n }\n\n case \"prepare_linkedin_post\": {\n const { text, task_id } = toolArgs as any;\n if (!text) return JSON.stringify({ error: \"text required\" });\n const action = await enqueueAction({\n integration: \"linkedin\", endpoint: \"create-post\",\n payload: { text },\n title: \"LinkedIn post\",\n preview: String(text).slice(0, 600),\n task_id, source_type: \"chat\", workspace: \"sales\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"LinkedIn post drafted and QUEUED FOR APPROVAL.\" });\n }\n\n case \"prepare_wordpress_post\": {\n const { title, content, status, task_id } = toolArgs as any;\n if (!title || !content) return JSON.stringify({ error: \"title + content required\" });\n const action = await enqueueAction({\n integration: \"wordpress\", endpoint: \"publish-post\",\n payload: { title, content, status: status === \"draft\" ? \"draft\" : \"publish\" },\n title: `WordPress: ${title}`,\n preview: `${title}\\n\\n${String(content).slice(0, 600)}`,\n task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"WordPress post drafted and QUEUED FOR APPROVAL.\" });\n }\n\n case \"create_task\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n try {\n const task = await createTask({\n title: a.title, description: a.description ?? null,\n priority: a.priority, workspace: a.workspace ?? \"marketing\",\n due_date: a.due_date, objective_id: a.objective_id,\n key_result_id: a.key_result_id, experiment_id: a.experiment_id,\n } as any);\n return JSON.stringify({ success: true, task });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create task\" });\n }\n }\n\n case \"schedule_calendar_event\": {\n const a = toolArgs as any;\n if (!a.title || !a.event_date) return JSON.stringify({ error: \"title + event_date (YYYY-MM-DD) required\" });\n try {\n const entry = await createCalendarEntry({\n title: a.title, event_date: a.event_date, event_time: a.event_time ?? null,\n workspace: a.workspace ?? \"marketing\", event_type: \"custom\",\n } as any);\n return JSON.stringify({ success: true, entry });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to schedule event\" });\n }\n }\n\n case \"list_objectives\": {\n const { workspace } = toolArgs as { workspace?: string };\n try {\n const objectives = await listObjectives(workspace ? { workspace } : {});\n return JSON.stringify({ objectives: objectives.map((o: any) => ({ id: o.id, title: o.title, status: o.status, progress: o.progress })) });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to list objectives\" });\n }\n }\n\n case \"create_campaign\": {\n const a = toolArgs as any;\n if (!a.name) return JSON.stringify({ error: \"name required\" });\n const workspace = a.workspace === \"sales\" ? \"sales\" : \"marketing\";\n try {\n // 1. OKR linkage — link an existing objective, else auto-create one (a campaign MUST tie to an OKR).\n let objectiveId = a.objective_id ? String(a.objective_id) : null;\n if (!objectiveId) {\n const now = new Date();\n const quarter = a.quarter || `Q${Math.floor(now.getMonth() / 3) + 1} ${now.getFullYear()}`;\n const obj = await createObjective({\n title: a.objective_title || `${a.name} objective`,\n description: a.goal ?? null,\n quarter,\n workspace,\n status: \"active\",\n } as any);\n objectiveId = obj.id;\n }\n // 2. ICE experiment — connects the campaign's impact to the objective so it gets measured.\n const experiment = await createExperiment({\n title: `${a.name} (campaign)`,\n hypothesis: a.goal ?? null,\n objective_id: objectiveId,\n funnel_stage: a.funnel_stage ?? null,\n channel: a.channel ?? null,\n status: \"idea\",\n impact: a.impact, confidence: a.confidence, ease: a.ease,\n } as any);\n // 3. The campaign itself (a campaign IS an agent).\n const campaign = await createAgent({\n name: a.name, workspace,\n description: a.goal ?? a.description ?? null,\n channel: a.channel ?? \"mixed\",\n goal: a.goal ?? null,\n audience: a.audience ?? null,\n status: a.status ?? \"active\",\n } as any);\n return JSON.stringify({\n success: true,\n campaign,\n linked: { objective_id: objectiveId, experiment_id: experiment.id },\n note: \"Campaign created and connected to an OKR objective + ICE experiment.\",\n });\n } catch (err: any) {\n return JSON.stringify({ error: err?.message ?? \"Failed to create campaign\" });\n }\n }\n\n case \"create_drive_folder\": {\n const a = toolArgs as any;\n if (!a.name) return JSON.stringify({ error: \"name required\" });\n const data = await runIntegrationWrite(\"drive\", \"createFolder\", { name: a.name });\n return JSON.stringify(data?.error ? data : { success: true, file: data });\n }\n\n case \"create_google_doc\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n const data = await runIntegrationWrite(\"docs\", \"create\", { title: a.title, content: a.content });\n return JSON.stringify(data?.error ? data : { success: true, doc: data });\n }\n\n case \"create_google_sheet\": {\n const a = toolArgs as any;\n if (!a.title) return JSON.stringify({ error: \"title required\" });\n const data = await runIntegrationWrite(\"sheets\", \"create\", { title: a.title });\n return JSON.stringify(data?.error ? data : { success: true, sheet: data });\n }\n\n case \"prepare_outlook_email\": {\n const a = toolArgs as any;\n if (!a.to || !a.subject) return JSON.stringify({ error: \"to + subject required\" });\n const action = await enqueueAction({\n integration: \"outlook\", endpoint: \"send\",\n payload: {\n message: {\n subject: a.subject,\n body: { contentType: \"Text\", content: a.body ?? \"\" },\n toRecipients: [{ emailAddress: { address: a.to } }],\n ...(a.cc ? { ccRecipients: [{ emailAddress: { address: a.cc } }] } : {}),\n },\n saveToSentItems: true,\n },\n title: `Outlook email: ${a.subject}`,\n preview: `To: ${a.to}\\nSubject: ${a.subject}\\n\\n${String(a.body ?? \"\").slice(0, 600)}`,\n task_id: a.task_id, source_type: \"chat\",\n });\n return JSON.stringify({ queued: true, status: \"prepared\", action_id: action?.id, note: \"Outlook email drafted and QUEUED FOR APPROVAL.\" });\n }\n\n default:\n return JSON.stringify({ error: `Unknown tool: ${toolName}` });\n }\n}\n\n// ─── Main chat handler ────────────────────────────────────────────────────────\n\nexport async function handleChat(req: Request): Promise<Response> {\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS });\n }\n\n if (req.method !== \"POST\") {\n return json({ error: \"Method not allowed\" }, 405);\n }\n\n let body: any;\n try {\n body = await req.json();\n } catch {\n return json({ error: \"Invalid JSON body\" }, 400);\n }\n\n const {\n message,\n agent_id,\n conversation_id: incomingConversationId,\n workspace,\n } = body as {\n message?: string;\n agent_id?: string;\n conversation_id?: string;\n workspace?: string;\n };\n\n if (!message || typeof message !== \"string\" || message.trim() === \"\") {\n return json({ error: \"message field is required and must be a non-empty string\" }, 400);\n }\n\n // Load LLM config\n const llmConfig = await getLLMConfig();\n\n // Load agent if specified\n let agent = null;\n if (agent_id) {\n agent = await getAgent(agent_id);\n }\n\n // Determine conversation ID\n const conversationId = incomingConversationId ?? crypto.randomUUID();\n\n // Load existing conversation history or start fresh\n let conversationHistory: ChatMessage[] = [];\n if (incomingConversationId) {\n const existing = await loadConversation(incomingConversationId);\n if (existing) {\n conversationHistory = existing.messages;\n }\n }\n\n // Add the new user message\n conversationHistory.push({ role: \"user\", content: message.trim() });\n\n // Build funnel-aware system prompt: inject live funnel state + top correlations\n // so the LLM always answers in causal/positional terms (TOFU/MOFU/BOFU).\n // Falls back silently if intelligence layer is unavailable.\n let funnelContext = \"\";\n try {\n const [summary, correlations] = await Promise.all([getFunnelSummary(), getCorrelations()]);\n const stageLines = summary.stages.map((s) =>\n ` ${s.stage} (${s.label}): health ${s.health_score}/100 · ${s.metrics.length} metrics · ${s.diagnosis}`\n ).join(\"\\n\");\n const bottleneck = summary.bottleneck ? `Bottleneck: ${summary.bottleneck.stage} — ${summary.bottleneck.reason}` : \"No bottleneck detected.\";\n const corrLines = correlations.slice(0, 5).map((c) =>\n ` - ${c.narrative} [${c.funnel_stage_from}→${c.funnel_stage_to}, ${c.strength}, lag ${c.lag_days}d, conf ${Math.round(c.confidence*100)}%]`\n ).join(\"\\n\");\n funnelContext = `\n\n## Current Funnel State (live, last 30d)\n${stageLines}\n${bottleneck}\nIntegrations connected: ${summary.total_integrations_connected} · Metrics observed: ${summary.total_metrics_observed}\n\n## Active Signal Correlations\n${corrLines || \" (no correlations yet — connect more integrations)\"}\n\n## Your Role\nYou are **Nexus** — a senior marketing & sales intelligence analyst embedded in the dashboard.\n- When a metric is mentioned, explain its funnel position (TOFU/MOFU/BOFU) and causal relationships\n- Proactively surface the highest-priority bottleneck in every response\n- Suggest specific actions tied to the integration data above\n- Use academic terminology: CAC, LTV, MQL, SQL, attribution, conversion velocity, message-market fit\n- End responses with a concrete next action when applicable\n`.trim();\n } catch { /* intelligence layer unavailable — proceed without context */ }\n\n // I3: agent role prompts. When agent.role is one of listener/topic/creator/analyst, prepend\n // the specialized prefix so the model behaves consistently with the role badge shown in the UI.\n const AGENT_ROLE_PROMPTS: Record<string, string> = {\n listener: `You are a Listener Agent. Your job:\n1. Monitor every incoming signal from Gmail, Outlook, and LinkedIn for buyer intent.\n2. Flag mentions of: pain points, competitor references, pricing questions, budget signals.\n3. Surface weekly patterns (\"3 leads this week mentioned X\").\n4. Err on the side of flagging — never miss a commercially-relevant signal.`,\n topic: `You are a Topic Agent. Your job:\n1. Take buyer pain points + intent signals (from the Listener Agent or Signals workspace).\n2. Generate content ideas: blog titles, LinkedIn angles, email subject lines.\n3. Score each idea by search-volume potential (Ahrefs), funnel-stage fit, estimated ICE impact.\n4. Output a prioritized content backlog formatted as: Title | Funnel Stage | AARRR Lever | Est. ICE Score.\n5. Link every output to an active OKR Key Result when possible.`,\n creator: `You are a Creator Agent. Your job:\n1. Take approved topic ideas and draft complete marketing assets.\n2. Match brand voice: direct, data-backed, B2B SaaS tone.\n3. Blog posts: H1/H2 structure, 800–1200 words, CTA at end.\n4. LinkedIn posts: hook in first line, insight in body, question as CTA, max 250 words.\n5. Email sequences: subject + 3-email sequence (awareness → consideration → CTA).\n6. Always cite one data point from the connected integrations as social proof.`,\n analyst: `You are an Analyst Agent. Your job:\n1. Analyze the funnel summary across all 14 integrations.\n2. Identify the single biggest bottleneck in the current TOFU/MOFU/BOFU funnel.\n3. Propose the top 3 ICE-scored experiments to address it.\n4. Track experiment outcomes and update affected OKR Key Results.\n5. Produce a weekly growth digest: wins, losses, signals, recommended experiments.`,\n };\n const rolePrompt = agent?.role && AGENT_ROLE_PROMPTS[agent.role] ? AGENT_ROLE_PROMPTS[agent.role] : \"\";\n\n const baseSystemPrompt: string =\n agent?.system_prompt?.trim()\n ? agent.system_prompt.trim()\n : DEFAULT_SYSTEM_PROMPT;\n const systemPrompt = [rolePrompt, baseSystemPrompt, EXECUTION_GUIDE, funnelContext]\n .filter((s) => s && s.length > 0)\n .join(\"\\n\\n\");\n\n // Determine tools — always give all tools; agents narrow by integration name via get_integration_data\n // Agent.tools stores integration names (e.g. [\"ga4\",\"gsc\"]), not tool function names,\n // so filtering by function name would always produce an empty list. Give all tools instead.\n const tools = buildTools();\n\n // Read temperature + frequency_penalty from app_config so Settings sliders are real, not cosmetic.\n let temperature: number | undefined = undefined;\n let frequencyPenalty: number | undefined = undefined;\n try {\n const cfgRows = await sqlite.execute(\"SELECT key, value FROM app_config WHERE key IN ('temperature','frequency_penalty')\");\n for (const row of cfgRows.rows) {\n const [k, v] = row as any[];\n const num = parseFloat(String(v).replace(/^\"|\"$/g, \"\"));\n if (!isNaN(num)) {\n if (k === \"temperature\") temperature = num;\n if (k === \"frequency_penalty\") frequencyPenalty = num;\n }\n }\n } catch { /* fall back to provider defaults */ }\n\n // First LLM call\n let llmResult = await callLLM(llmConfig, conversationHistory, {\n systemPrompt,\n tools,\n maxTokens: 4096,\n temperature,\n frequencyPenalty,\n });\n\n const allToolCalls: any[] = [];\n\n // Handle tool use — bounded multi-round loop (was a single round, which truncated\n // any plan needing chained tool calls). Cap rounds to avoid runaway cost/latency.\n let toolRounds = 0;\n const MAX_TOOL_ROUNDS = 5;\n while (llmResult.tool_calls && llmResult.tool_calls.length > 0 && toolRounds < MAX_TOOL_ROUNDS) {\n toolRounds++;\n // Record the assistant message with tool calls.\n // reasoning_content must be echoed back for DeepSeek thinking-mode multi-turn.\n const assistantMsg: any = {\n role: \"assistant\",\n content: llmResult.content || \"\",\n tool_calls: llmResult.tool_calls,\n };\n if ((llmResult as any).reasoning_content) {\n assistantMsg.reasoning_content = (llmResult as any).reasoning_content;\n }\n conversationHistory.push(assistantMsg);\n\n // Execute each tool call and collect results\n const toolResults: Array<{ tool_call_id: string; role: \"tool\"; content: string }> = [];\n\n for (const toolCall of llmResult.tool_calls) {\n allToolCalls.push(toolCall);\n\n let args: Record<string, any> = {};\n try {\n args =\n typeof toolCall.function.arguments === \"string\"\n ? JSON.parse(toolCall.function.arguments)\n : toolCall.function.arguments ?? {};\n } catch {\n args = {};\n }\n\n const toolResult = await executeTool(toolCall.function.name, args, llmConfig);\n\n toolResults.push({\n tool_call_id: toolCall.id,\n role: \"tool\",\n content: toolResult,\n });\n }\n\n // Add tool results to history\n for (const tr of toolResults) {\n conversationHistory.push(tr as any);\n }\n\n // Second LLM call with tool results\n llmResult = await callLLM(llmConfig, conversationHistory, {\n systemPrompt,\n tools,\n maxTokens: 4096,\n temperature,\n frequencyPenalty,\n });\n }\n\n // Add final assistant response to history\n conversationHistory.push({\n role: \"assistant\",\n content: llmResult.content,\n });\n\n // Persist conversation\n await saveConversation(\n conversationId,\n conversationHistory,\n agent_id,\n workspace\n );\n\n return json({\n response: llmResult.content,\n conversation_id: conversationId,\n tool_calls: allToolCalls.length > 0 ? allToolCalls : undefined,\n agent_id: agent_id ?? undefined,\n });\n}\n\n// ─── Streaming chat handler (SSE) ────────────────────────────────────────────\n\nexport async function handleChatStream(req: Request): Promise<Response> {\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS });\n }\n if (req.method !== \"POST\") {\n return new Response(JSON.stringify({ error: \"Method not allowed\" }), { status: 405, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n let body: any;\n try { body = await req.json(); } catch {\n return new Response(JSON.stringify({ error: \"Invalid JSON body\" }), { status: 400, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n const { message, agent_id, conversation_id: incomingConversationId, workspace } = body as {\n message?: string; agent_id?: string; conversation_id?: string; workspace?: string;\n };\n\n if (!message || typeof message !== \"string\" || message.trim() === \"\") {\n return new Response(JSON.stringify({ error: \"message field is required\" }), { status: 400, headers: { \"Content-Type\": \"application/json\", ...CORS } });\n }\n\n const llmConfig = await getLLMConfig();\n let agent = null;\n if (agent_id) { agent = await getAgent(agent_id); }\n\n const conversationId = incomingConversationId ?? crypto.randomUUID();\n let conversationHistory: ChatMessage[] = [];\n if (incomingConversationId) {\n const existing = await loadConversation(incomingConversationId);\n if (existing) conversationHistory = existing.messages;\n }\n conversationHistory.push({ role: \"user\", content: message.trim() });\n\n const baseSystemPrompt: string = agent?.system_prompt?.trim() ? agent.system_prompt.trim() :\n \"You are an AI assistant for AIHive Command Center, a marketing and sales platform.\";\n\n const encoder = new TextEncoder();\n const stream = new ReadableStream({\n async start(controller) {\n const send = (data: object) => {\n controller.enqueue(encoder.encode(`data: ${JSON.stringify(data)}\\n\\n`));\n };\n\n let fullResponse = \"\";\n try {\n const gen = callLLMStream(llmConfig, conversationHistory, {\n systemPrompt: baseSystemPrompt, maxTokens: 4096,\n });\n for await (const token of gen) {\n if (token) {\n fullResponse += token;\n send({ delta: token, done: false });\n }\n }\n // Persist the full conversation\n conversationHistory.push({ role: \"assistant\", content: fullResponse });\n await saveConversation(conversationId, conversationHistory, agent_id, workspace);\n send({ delta: \"\", done: true, response: fullResponse, conversation_id: conversationId });\n } catch (err: any) {\n send({ delta: \"\", done: true, error: err?.message ?? \"Stream error\", conversation_id: conversationId });\n } finally {\n controller.close();\n }\n },\n });\n\n return new Response(stream, {\n headers: {\n \"Content-Type\": \"text/event-stream\",\n \"Cache-Control\": \"no-cache\",\n \"Connection\": \"keep-alive\",\n ...CORS,\n },\n });\n}\n\n// ─── Conversation history CRUD ────────────────────────────────────────────────\n\nexport async function listConversations(): Promise<any[]> {\n try {\n const result = await sqlite.execute(\n `SELECT id, workspace, agent_id, created_at, updated_at, messages\n FROM conversations ORDER BY updated_at DESC LIMIT 50`\n );\n return result.rows.map((row) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = row[i];\n // Extract first user message as title\n let firstMsg = \"\";\n try {\n const msgs: ChatMessage[] = JSON.parse(String(obj.messages || \"[]\"));\n const first = msgs.find((m) => m.role === \"user\");\n firstMsg = first ? String(first.content).slice(0, 60) : \"\";\n } catch {}\n return {\n id: obj.id,\n workspace: obj.workspace,\n agent_id: obj.agent_id,\n created_at: obj.created_at,\n updated_at: obj.updated_at,\n title: firstMsg || \"(no messages)\",\n };\n });\n } catch { return []; }\n}\n\nexport async function getConversationById(id: string): Promise<any> {\n try {\n const result = await sqlite.execute(\n \"SELECT id, workspace, agent_id, created_at, updated_at, messages FROM conversations WHERE id = ?\",\n [id]\n );\n if (!result.rows?.length) return null;\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = result.rows[0][i];\n let messages: ChatMessage[] = [];\n try { messages = JSON.parse(String(obj.messages || \"[]\")); } catch {}\n return { id: obj.id, workspace: obj.workspace, agent_id: obj.agent_id,\n created_at: obj.created_at, updated_at: obj.updated_at, messages };\n } catch { return null; }\n}\n\nexport async function deleteConversation(id: string): Promise<boolean> {\n try {\n await sqlite.execute(\"DELETE FROM conversations WHERE id = ?\", [id]);\n return true;\n } catch { return false; }\n}\n",[],[365,869],"bb0eff04-5b4d-11f1-a9c0-ee650bb23af1","calendar_unified.ts","98d397f8-589e-11f1-a990-ee650bb23af1",511,["D",1780052483377],"// Unified Calendar — merges six event sources (tasks, experiments, objectives,\n// key results, Google Calendar live, custom entries) into a single sorted feed.\n// Every source is wrapped in try/catch so a missing table or down API never\n// blocks the rest of the calendar from loading.\nimport { sqlite } from \"../db.ts\";\nimport { getCredentials } from \"./integrations/_shared.ts\";\nimport { callLLM, getLLMConfig } from \"./llm.ts\";\n\n// ─── Types ───────────────────────────────────────────────────────────────────\n\nexport type CalendarSource =\n | \"tasks\" | \"experiments\" | \"objectives\"\n | \"key_results\" | \"google_calendar\" | \"outlook_calendar\" | \"custom\" | \"nsm\" | \"actions\";\n\nexport interface CalendarEvent {\n id: string; // source-prefixed: \"task:abc\", \"exp_start:xyz\", ...\n source: CalendarSource;\n title: string;\n description?: string;\n date: string; // YYYY-MM-DD (start)\n time?: string; // HH:MM (start) — omitted if all_day\n end_date?: string; // YYYY-MM-DD (end)\n end_time?: string; // HH:MM (end)\n all_day: boolean;\n color: string;\n workspace?: string;\n event_type?: string;\n linked_url?: string; // deep-link back into the originating module\n meta?: Record<string, any>;\n}\n\n// ─── Color / helper maps ─────────────────────────────────────────────────────\n\nconst SOURCE_COLORS: Record<CalendarSource, string> = {\n tasks: \"#3B82F6\", // blue\n experiments: \"#A855F7\", // purple\n objectives: \"#10B981\", // emerald\n key_results: \"#F59E0B\", // amber\n google_calendar: \"#EF4444\", // red\n outlook_calendar:\"#0078D4\", // Outlook blue\n custom: \"#6B7280\", // slate\n nsm: \"#EC4899\", // pink — North Star Metric deadline\n actions: \"#14B8A6\", // teal — AI/agent action items (action_queue)\n};\n\nexport function colorFor(source: CalendarSource): string {\n return SOURCE_COLORS[source] ?? \"#6B7280\";\n}\n\nexport function quarterToEndDate(quarter: string): string | null {\n // Accepts \"2026-Q1\", \"Q1-2026\", \"Q1 2026\", \"Q1/2026\".\n const m = String(quarter).match(/Q?\\s*([1-4]).{0,3}?(\\d{4})|(\\d{4}).{0,3}?Q?\\s*([1-4])/i);\n if (!m) return null;\n const q = parseInt(m[1] || m[4], 10);\n const y = parseInt(m[2] || m[3], 10);\n if (!q || !y) return null;\n switch (q) {\n case 1: return `${y}-03-31`;\n case 2: return `${y}-06-30`;\n case 3: return `${y}-09-30`;\n case 4: return `${y}-12-31`;\n default: return null;\n }\n}\n\nfunction todayISO(): string {\n const d = new Date(); d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\nfunction mondayISO(): string {\n const d = new Date();\n const dow = (d.getDay() + 6) % 7;\n d.setDate(d.getDate() - dow);\n d.setHours(0, 0, 0, 0);\n return d.toISOString().slice(0, 10);\n}\n\nfunction addDaysISO(iso: string, days: number): string {\n const d = new Date(iso + \"T00:00:00Z\");\n d.setUTCDate(d.getUTCDate() + days);\n return d.toISOString().slice(0, 10);\n}\n\nfunction isoDateOnly(input: any): string | null {\n if (!input) return null;\n const s = String(input);\n // Already YYYY-MM-DD\n if (/^\\d{4}-\\d{2}-\\d{2}$/.test(s)) return s;\n const d = new Date(s);\n if (isNaN(d.getTime())) return null;\n return d.toISOString().slice(0, 10);\n}\n\nfunction isoTime(input: any): string | undefined {\n if (!input) return undefined;\n const s = String(input);\n // HH:MM or HH:MM:SS\n if (/^\\d{2}:\\d{2}(:\\d{2})?$/.test(s)) return s.slice(0, 5);\n const d = new Date(s);\n if (isNaN(d.getTime())) return undefined;\n return d.toISOString().slice(11, 16);\n}\n\nfunction inRange(date: string | null, from: string, to: string): boolean {\n if (!date) return false;\n return date >= from && date <= to;\n}\n\n// ─── Source fetchers ─────────────────────────────────────────────────────────\n\nasync function fetchTasks(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n // tasks table may or may not have a due_date column — query the schema first.\n const cols = await sqlite.execute(\"PRAGMA table_info(tasks)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n const hasDue = colNames.has(\"due_date\");\n if (!hasDue) return [];\n\n const where = [\"due_date IS NOT NULL\", \"date(due_date) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, workspace, status, due_date FROM tasks WHERE ${where.join(\" AND \")} ORDER BY due_date ASC`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const due = isoDateOnly(row[4]);\n if (!due) continue;\n events.push({\n id: `task:${row[0]}`,\n source: \"tasks\",\n title: String(row[1] ?? \"Task\"),\n date: due,\n all_day: true,\n color: colorFor(\"tasks\"),\n workspace: row[2] ?? undefined,\n event_type: \"task_due\",\n linked_url: `/tasks/${row[0]}`,\n meta: { status: row[3] },\n });\n }\n } catch { /* missing table or column → skip */ }\n return events;\n}\n\n// Action items = the action_queue (what the AI/agents prepared, approved, executed). ALL statuses\n// are shown (prepared/approved/executing/done/failed/rejected) — not just completed ones.\nasync function fetchActionItems(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(action_queue)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const where = [\"date(created_at) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace && colNames.has(\"workspace\")) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, integration, endpoint, status, workspace, created_at\n FROM action_queue WHERE ${where.join(\" AND \")} ORDER BY created_at ASC`,\n args,\n );\n for (const row of (r.rows || []) as any[]) {\n const date = isoDateOnly(row[6]);\n if (!date) continue;\n const status = String(row[4] ?? \"prepared\");\n events.push({\n id: `action:${row[0]}`,\n source: \"actions\",\n title: String(row[1] || `${row[2] ?? \"\"} ${row[3] ?? \"\"}`.trim() || \"Action item\"),\n date,\n all_day: true,\n color: colorFor(\"actions\"),\n workspace: row[5] ?? undefined,\n event_type: `action_${status}`,\n linked_url: `/approvals`,\n meta: { status, integration: row[2], endpoint: row[3] },\n });\n }\n } catch { /* missing action_queue → skip */ }\n return events;\n}\n\nasync function fetchExperiments(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n // Probe schema for compatibility — experiments table may have name/title, start_date/end_date.\n const cols = await sqlite.execute(\"PRAGMA table_info(experiments)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"name\") ? \"name\" : colNames.has(\"title\") ? \"title\" : null;\n if (!titleCol || !colNames.has(\"start_date\") || !colNames.has(\"end_date\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const args: any[] = [from, to, from, to];\n if (wsFilter) args.push(workspace);\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, start_date, end_date${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}${colNames.has(\"status\") ? \", status\" : \", NULL\"}\n FROM experiments\n WHERE (date(start_date) BETWEEN date(?) AND date(?))\n OR (date(end_date) BETWEEN date(?) AND date(?))\n ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const id = row[0];\n const title = String(row[1] ?? \"Experiment\");\n const start = isoDateOnly(row[2]);\n const end = isoDateOnly(row[3]);\n const ws = row[4] ?? undefined;\n const status = row[5] ?? undefined;\n if (inRange(start, from, to)) {\n events.push({\n id: `exp_start:${id}`,\n source: \"experiments\",\n title: `Start: ${title}`,\n date: start!,\n all_day: true,\n color: colorFor(\"experiments\"),\n workspace: ws,\n event_type: \"experiment_start\",\n linked_url: `/experiments/${id}`,\n meta: { status },\n });\n }\n if (inRange(end, from, to)) {\n events.push({\n id: `exp_end:${id}`,\n source: \"experiments\",\n title: `End: ${title}`,\n date: end!,\n all_day: true,\n color: colorFor(\"experiments\"),\n workspace: ws,\n event_type: \"experiment_end\",\n linked_url: `/experiments/${id}`,\n meta: { status },\n });\n }\n }\n } catch { /* missing experiments table → skip */ }\n return events;\n}\n\nasync function fetchObjectives(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(objectives)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"title\") ? \"title\" : colNames.has(\"name\") ? \"name\" : null;\n if (!titleCol || !colNames.has(\"quarter\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const args: any[] = [];\n if (wsFilter) args.push(workspace);\n const statusFilter = colNames.has(\"status\") ? \"AND status = 'active'\" : \"\";\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, quarter${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}\n FROM objectives WHERE 1=1 ${statusFilter} ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const quarterEnd = quarterToEndDate(String(row[2] ?? \"\"));\n if (!quarterEnd) continue;\n if (!inRange(quarterEnd, from, to)) continue;\n events.push({\n id: `obj:${row[0]}`,\n source: \"objectives\",\n title: `Objective due: ${row[1]}`,\n date: quarterEnd,\n all_day: true,\n color: colorFor(\"objectives\"),\n workspace: row[3] ?? undefined,\n event_type: \"objective_due\",\n linked_url: `/okrs/${row[0]}`,\n meta: { quarter: row[2] },\n });\n }\n } catch { /* missing objectives table */ }\n return events;\n}\n\nasync function fetchKeyResults(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(key_results)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n if (colNames.size === 0) return [];\n const titleCol = colNames.has(\"title\") ? \"title\" : colNames.has(\"name\") ? \"name\" : null;\n if (!titleCol || !colNames.has(\"updated_at\")) return [];\n const wsFilter = workspace && colNames.has(\"workspace\") ? \"AND workspace = ?\" : \"\";\n const progFilter = colNames.has(\"progress\") ? \"AND progress < 100\" : \"\";\n const args: any[] = [from, to];\n if (wsFilter) args.push(workspace);\n const r = await sqlite.execute(\n `SELECT id, ${titleCol}, updated_at${colNames.has(\"progress\") ? \", progress\" : \", 0\"}${colNames.has(\"workspace\") ? \", workspace\" : \", NULL\"}\n FROM key_results\n WHERE date(updated_at) BETWEEN date(?) AND date(?)\n ${progFilter} ${wsFilter}`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const updated = isoDateOnly(row[2]);\n if (!updated) continue;\n events.push({\n id: `kr:${row[0]}`,\n source: \"key_results\",\n title: `KR update: ${row[1]}`,\n date: updated,\n all_day: true,\n color: colorFor(\"key_results\"),\n workspace: row[4] ?? undefined,\n event_type: \"kr_update\",\n linked_url: `/okrs?kr=${row[0]}`,\n meta: { progress: Number(row[3]) || 0 },\n });\n }\n } catch { /* missing key_results table */ }\n return events;\n}\n\nasync function fetchGoogleCalendar(from: string, to: string): Promise<CalendarEvent[]> {\n try {\n // Calendar inherits from `google` OAuth credentials; try `calendar` first.\n let creds = await getCredentials(\"calendar\");\n if (!creds?.access_token) creds = await getCredentials(\"google\");\n if (!creds?.access_token) return [];\n\n const timeMin = `${from}T00:00:00Z`;\n const timeMax = `${to}T23:59:59Z`;\n const url = `https://www.googleapis.com/calendar/v3/calendars/primary/events?` +\n `timeMin=${encodeURIComponent(timeMin)}` +\n `&timeMax=${encodeURIComponent(timeMax)}` +\n `&singleEvents=true&orderBy=startTime&maxResults=250`;\n const res = await fetch(url, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return [];\n const data = await res.json().catch(() => ({}));\n const items = Array.isArray(data?.items) ? data.items : [];\n const out: CalendarEvent[] = [];\n for (const ev of items) {\n const startObj = ev.start || {};\n const endObj = ev.end || {};\n const isAllDay = !!startObj.date && !startObj.dateTime;\n const startISO = startObj.dateTime || startObj.date;\n const endISO = endObj.dateTime || endObj.date;\n const date = isoDateOnly(startISO);\n if (!date) continue;\n out.push({\n id: `gcal:${ev.id}`,\n source: \"google_calendar\",\n title: String(ev.summary || \"(no title)\"),\n description: ev.description ? String(ev.description) : undefined,\n date,\n time: isAllDay ? undefined : isoTime(startISO),\n end_date: isoDateOnly(endISO) ?? undefined,\n end_time: isAllDay ? undefined : isoTime(endISO),\n all_day: isAllDay,\n color: colorFor(\"google_calendar\"),\n event_type: \"google_event\",\n linked_url: ev.htmlLink ? String(ev.htmlLink) : undefined,\n meta: { hangoutLink: ev.hangoutLink, location: ev.location, attendees: (ev.attendees || []).length },\n });\n }\n return out;\n } catch {\n return []; // Google down or token expired — entire calendar still loads.\n }\n}\n\nasync function fetchOutlookCalendar(from: string, to: string): Promise<CalendarEvent[]> {\n try {\n // Shares the central `outlook` Microsoft Graph token (same as the Outlook integration).\n const creds = await getCredentials(\"outlook\");\n if (!creds?.access_token) return [];\n const startISO = `${from}T00:00:00Z`;\n const endISO = `${to}T23:59:59Z`;\n const url = `https://graph.microsoft.com/v1.0/me/calendarView?` +\n `startDateTime=${encodeURIComponent(startISO)}&endDateTime=${encodeURIComponent(endISO)}` +\n `&$top=250&$orderby=start/dateTime` +\n `&$select=subject,start,end,isAllDay,webLink,location,organizer`;\n const res = await fetch(url, {\n headers: { Authorization: `Bearer ${creds.access_token}`, Prefer: 'outlook.timezone=\"UTC\"' },\n });\n if (!res.ok) return [];\n const data = await res.json().catch(() => ({}));\n const items = Array.isArray(data?.value) ? data.value : [];\n const out: CalendarEvent[] = [];\n for (const ev of items) {\n const startStr = ev.start?.dateTime || ev.start?.date;\n const endStr = ev.end?.dateTime || ev.end?.date;\n const date = isoDateOnly(startStr);\n if (!date) continue;\n const isAllDay = !!ev.isAllDay;\n out.push({\n id: `ocal:${ev.id}`,\n source: \"outlook_calendar\",\n title: String(ev.subject || \"(no title)\"),\n date,\n time: isAllDay ? undefined : isoTime(startStr),\n end_date: isoDateOnly(endStr) ?? undefined,\n end_time: isAllDay ? undefined : isoTime(endStr),\n all_day: isAllDay,\n color: colorFor(\"outlook_calendar\"),\n event_type: \"outlook_event\",\n linked_url: ev.webLink ? String(ev.webLink) : undefined,\n meta: { location: ev.location?.displayName, organizer: ev.organizer?.emailAddress?.name },\n });\n }\n return out;\n } catch {\n return []; // Outlook down or token lacks Calendars.Read — rest of the calendar still loads.\n }\n}\n\nasync function fetchNSM(from: string, to: string): Promise<CalendarEvent[]> {\n // Surface the North Star Metric deadline as a single calendar event so the team\n // can see \"KPI target due\" on the calendar. NSM is in app_config; we read the\n // most recent target_date.\n const events: CalendarEvent[] = [];\n try {\n const r = await sqlite.execute(\n `SELECT id, metric_name, target_value, target_date, current_value, unit\n FROM north_star WHERE target_date IS NOT NULL ORDER BY set_at DESC LIMIT 1`\n );\n const rows = (r.rows || []) as any[];\n if (rows.length === 0) return [];\n const row = rows[0];\n const date = isoDateOnly(row[3]);\n if (!date || !inRange(date, from, to)) return [];\n const name = String(row[1] ?? \"North Star Metric\");\n const target = row[2];\n const unit = row[5] ? String(row[5]) : \"\";\n events.push({\n id: `nsm:${row[0]}`,\n source: \"nsm\",\n title: `NSM target due: ${name} → ${target}${unit ? \" \" + unit : \"\"}`,\n date,\n all_day: true,\n color: colorFor(\"nsm\"),\n event_type: \"nsm_target\",\n linked_url: `/command`,\n meta: { current: row[4], target },\n });\n } catch { /* nsm_history missing → skip */ }\n return events;\n}\n\nasync function fetchCustom(from: string, to: string, workspace?: string): Promise<CalendarEvent[]> {\n const events: CalendarEvent[] = [];\n try {\n const where = [\"date(event_date) BETWEEN date(?) AND date(?)\"];\n const args: any[] = [from, to];\n if (workspace) { where.push(\"workspace = ?\"); args.push(workspace); }\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence\n FROM calendar_entries WHERE ${where.join(\" AND \")} ORDER BY event_date ASC, event_time ASC`,\n args\n );\n for (const row of (r.rows || []) as any[]) {\n const date = isoDateOnly(row[3]);\n if (!date) continue;\n const allDay = Number(row[10]) === 1;\n events.push({\n id: `cal:${row[0]}`,\n source: \"custom\",\n title: String(row[1] ?? \"Event\"),\n description: row[2] ?? undefined,\n date,\n time: allDay ? undefined : isoTime(row[4]),\n end_date: isoDateOnly(row[5]) ?? undefined,\n end_time: allDay ? undefined : isoTime(row[6]),\n all_day: allDay,\n color: row[9] ? String(row[9]) : colorFor(\"custom\"),\n workspace: row[8] ?? undefined,\n event_type: row[7] ? String(row[7]) : \"custom\",\n meta: { recurrence: row[11] ?? null },\n });\n }\n } catch { /* missing calendar_entries → skip */ }\n return events;\n}\n\n// ─── Unified entry point ─────────────────────────────────────────────────────\n\nexport async function getUnifiedCalendar(opts: {\n from: string;\n to: string;\n workspace?: string;\n sources?: CalendarSource[];\n}): Promise<CalendarEvent[]> {\n const from = opts.from;\n const to = opts.to;\n const all: CalendarSource[] = [\"tasks\", \"experiments\", \"objectives\", \"key_results\", \"google_calendar\", \"outlook_calendar\", \"custom\", \"nsm\", \"actions\"];\n const wanted = new Set(opts.sources?.length ? opts.sources : all);\n\n const tasks: Promise<CalendarEvent[]>[] = [];\n if (wanted.has(\"actions\")) tasks.push(fetchActionItems(from, to, opts.workspace));\n if (wanted.has(\"tasks\")) tasks.push(fetchTasks(from, to, opts.workspace));\n if (wanted.has(\"experiments\")) tasks.push(fetchExperiments(from, to, opts.workspace));\n if (wanted.has(\"objectives\")) tasks.push(fetchObjectives(from, to, opts.workspace));\n if (wanted.has(\"key_results\")) tasks.push(fetchKeyResults(from, to, opts.workspace));\n if (wanted.has(\"google_calendar\")) tasks.push(fetchGoogleCalendar(from, to));\n if (wanted.has(\"outlook_calendar\")) tasks.push(fetchOutlookCalendar(from, to));\n if (wanted.has(\"custom\")) tasks.push(fetchCustom(from, to, opts.workspace));\n if (wanted.has(\"nsm\")) tasks.push(fetchNSM(from, to));\n\n const results = await Promise.allSettled(tasks);\n const merged: CalendarEvent[] = [];\n for (const r of results) {\n if (r.status === \"fulfilled\" && Array.isArray(r.value)) merged.push(...r.value);\n }\n\n merged.sort((a, b) => {\n if (a.date !== b.date) return a.date < b.date ? -1 : 1;\n const at = a.time || \"00:00\";\n const bt = b.time || \"00:00\";\n return at < bt ? -1 : at > bt ? 1 : 0;\n });\n return merged;\n}\n\nexport async function getToday(opts?: { workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const t = todayISO();\n return getUnifiedCalendar({ from: t, to: t, workspace: opts?.workspace, sources: opts?.sources });\n}\n\nexport async function getWeek(opts?: { workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const start = mondayISO();\n const end = addDaysISO(start, 6);\n return getUnifiedCalendar({ from: start, to: end, workspace: opts?.workspace, sources: opts?.sources });\n}\n\nexport async function getUpcoming(opts?: { days?: number; workspace?: string; sources?: CalendarSource[] }): Promise<CalendarEvent[]> {\n const days = opts?.days ?? 14;\n const start = todayISO();\n const end = addDaysISO(start, days);\n return getUnifiedCalendar({ from: start, to: end, workspace: opts?.workspace, sources: opts?.sources });\n}\n\n// ─── CRUD on calendar_entries ────────────────────────────────────────────────\n\nexport interface CalendarEntry {\n id: string;\n title: string;\n description: string | null;\n event_date: string;\n event_time: string | null;\n end_date: string | null;\n end_time: string | null;\n event_type: string | null;\n workspace: string | null;\n color: string | null;\n all_day: number;\n recurrence: string | null;\n created_at: string;\n}\n\nfunction rowToEntry(columns: string[], row: any[]): CalendarEntry {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n return {\n id: String(obj.id),\n title: String(obj.title ?? \"\"),\n description: obj.description ?? null,\n event_date: String(obj.event_date),\n event_time: obj.event_time ?? null,\n end_date: obj.end_date ?? null,\n end_time: obj.end_time ?? null,\n event_type: obj.event_type ?? null,\n workspace: obj.workspace ?? null,\n color: obj.color ?? null,\n all_day: Number(obj.all_day) || 0,\n recurrence: obj.recurrence ?? null,\n created_at: String(obj.created_at ?? new Date().toISOString()),\n };\n}\n\nexport async function createCalendarEntry(data: Partial<CalendarEntry>): Promise<CalendarEntry> {\n // QA fix: was passing a UUID string into INTEGER PRIMARY KEY AUTOINCREMENT —\n // SQLITE_MISMATCH datatype mismatch crashed every entry create. Omit id and use\n // last_insert_rowid() to read back the assigned numeric id.\n const now = new Date().toISOString();\n const eventDate = isoDateOnly(data.event_date) ?? todayISO();\n await sqlite.execute(\n `INSERT INTO calendar_entries\n (title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [\n data.title ?? \"Untitled\",\n data.description ?? null,\n eventDate,\n data.event_time ?? null,\n isoDateOnly(data.end_date) ?? null,\n data.end_time ?? null,\n data.event_type ?? null,\n data.workspace ?? null,\n data.color ?? null,\n data.all_day ? 1 : 0,\n data.recurrence ?? null,\n now,\n ]\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n const id = String((idRow.rows[0] as any[])[0]);\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) {\n return {\n id, title: data.title ?? \"Untitled\", description: data.description ?? null,\n event_date: eventDate, event_time: data.event_time ?? null,\n end_date: isoDateOnly(data.end_date) ?? null, end_time: data.end_time ?? null,\n event_type: data.event_type ?? null, workspace: data.workspace ?? null,\n color: data.color ?? null, all_day: data.all_day ? 1 : 0,\n recurrence: data.recurrence ?? null, created_at: now,\n };\n }\n return rowToEntry(r.columns, r.rows[0] as any[]);\n}\n\nexport async function updateCalendarEntry(id: string, data: Partial<CalendarEntry>): Promise<CalendarEntry | null> {\n const allowed: Array<keyof CalendarEntry> = [\n \"title\", \"description\", \"event_date\", \"event_time\", \"end_date\", \"end_time\",\n \"event_type\", \"workspace\", \"color\", \"all_day\", \"recurrence\",\n ];\n const sets: string[] = [];\n const args: any[] = [];\n for (const k of allowed) {\n if ((data as any)[k] === undefined) continue;\n sets.push(`${k} = ?`);\n let v: any = (data as any)[k];\n if (k === \"event_date\" || k === \"end_date\") v = isoDateOnly(v);\n if (k === \"all_day\") v = v ? 1 : 0;\n args.push(v);\n }\n if (sets.length === 0) {\n const cur = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!cur.rows?.length) return null;\n return rowToEntry(cur.columns, cur.rows[0] as any[]);\n }\n args.push(id);\n await sqlite.execute(`UPDATE calendar_entries SET ${sets.join(\", \")} WHERE id = ?`, args);\n const r = await sqlite.execute(\n `SELECT id, title, description, event_date, event_time, end_date, end_time,\n event_type, workspace, color, all_day, recurrence, created_at\n FROM calendar_entries WHERE id = ?`,\n [id]\n );\n if (!r.rows?.length) return null;\n return rowToEntry(r.columns, r.rows[0] as any[]);\n}\n\nexport async function deleteCalendarEntry(id: string): Promise<boolean> {\n await sqlite.execute(\"DELETE FROM calendar_entries WHERE id = ?\", [id]);\n return true;\n}\n\n// Clone a task into a calendar_entries row — lets users \"pin\" a task to the calendar\n// independently of due_date changes on the source.\nexport async function cloneTaskAsEntry(task_id: string): Promise<CalendarEntry | null> {\n try {\n const cols = await sqlite.execute(\"PRAGMA table_info(tasks)\");\n const colNames = new Set((cols.rows || []).map((r: any) => String(Array.isArray(r) ? r[1] : r.name)));\n const hasDue = colNames.has(\"due_date\");\n const hasDesc = colNames.has(\"description\");\n const selCols = [\"id\", \"title\", \"workspace\"];\n if (hasDue) selCols.push(\"due_date\"); else selCols.push(\"NULL AS due_date\");\n if (hasDesc) selCols.push(\"description\"); else selCols.push(\"NULL AS description\");\n const r = await sqlite.execute(\n `SELECT ${selCols.join(\", \")} FROM tasks WHERE id = ?`,\n [task_id]\n );\n if (!r.rows?.length) return null;\n const row = r.rows[0] as any[];\n const due = isoDateOnly(row[3]) ?? todayISO();\n return await createCalendarEntry({\n title: String(row[1] ?? \"Task\"),\n description: row[4] ?? null,\n event_date: due,\n workspace: row[2] ?? null,\n event_type: \"task_clone\",\n all_day: 1,\n });\n } catch { return null; }\n}\n\n// ─── LLM agenda briefing ─────────────────────────────────────────────────────\n\nfunction withTimeout<T>(p: Promise<T>, ms: number): Promise<T | null> {\n return new Promise<T | null>((resolve) => {\n let done = false;\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null); } }, ms);\n p.then(\n (v) => { if (!done) { done = true; clearTimeout(timer); resolve(v); } },\n () => { if (!done) { done = true; clearTimeout(timer); resolve(null); } },\n );\n });\n}\n\nfunction deterministicSummary(events: CalendarEvent[], date: string): string {\n if (events.length === 0) return `${date}: no scheduled events. Use the open block to ship one needle-mover.`;\n const counts = new Map<string, number>();\n for (const e of events) counts.set(e.source, (counts.get(e.source) || 0) + 1);\n const parts = Array.from(counts.entries()).map(([s, n]) => `${n} ${s.replace(\"_\", \" \")}`);\n const headline = events[0];\n return `${date}: ${events.length} item${events.length > 1 ? \"s\" : \"\"} (${parts.join(\", \")}). First up: ${headline.title}${headline.time ? ` at ${headline.time}` : \"\"}.`;\n}\n\nexport async function getCalendarSummary(date: string): Promise<{ date: string; summary: string; event_count: number }> {\n const day = isoDateOnly(date) ?? todayISO();\n const events = await getUnifiedCalendar({ from: day, to: day });\n const fallback = deterministicSummary(events, day);\n\n try {\n const cfg = await getLLMConfig();\n if (!cfg?.api_key) return { date: day, summary: fallback, event_count: events.length };\n\n const compact = events.map((e) => ({\n title: e.title, source: e.source, time: e.time, all_day: e.all_day, type: e.event_type,\n }));\n const prompt = [\n \"You are an executive assistant. Produce a 2-3 sentence agenda briefing for the day.\",\n \"Highlight time-sensitive items first, then context-switching cost, then one focus recommendation.\",\n \"Plain prose, no lists, under 350 chars total.\",\n \"\",\n `DATE: ${day}`,\n `EVENTS: ${JSON.stringify(compact)}`,\n ].join(\"\\n\");\n const llmCall = callLLM(cfg, [{ role: \"user\", content: prompt }], {\n maxTokens: 300,\n temperature: 0.3,\n systemPrompt: \"You write punchy executive briefings.\",\n });\n const result = await withTimeout(llmCall, 10000) as { content?: string } | null;\n const summary = result?.content?.trim() || fallback;\n return { date: day, summary, event_count: events.length };\n } catch {\n return { date: day, summary: fallback, event_count: events.length };\n }\n}\n",[],[365,878],"23580344-5d6a-11f1-8689-1607ee4eb77e","aio.ts","31d1fe3c-5821-11f1-8cb6-ee650bb23af1",634,["D",1780284586582],"// AIO math layer — Phase 5 / Group D:\n// D1: cohort survival LTV (customer_cohorts table)\n// D3: lead engagement velocity (lead_state_changes table)\n// D5: channel attribution (utm_conversions table)\n// Kept separate from intelligence.ts so the math engine file stays focused on funnel/correlations.\nimport { sqlite } from \"../db.ts\";\n\n// ─── D1: COHORT-SURVIVAL LTV ─────────────────────────────────────────────────\n// retention_curve = [m1, m2, m3, ...] retained counts. LTV per customer is the geometric\n// sum of monthly ARPU weighted by the retention rate at each month:\n// LTV = ARPU × Σ (retained_m / cohort_size)\n// Falls back to ARPU / churn_rate when only a single retention value is provided.\n\nexport interface CustomerCohort {\n id: string;\n cohort_label: string;\n cohort_size: number;\n arpu: number;\n retention_curve: number[];\n created_at: string;\n}\n\nfunction rowToCohort(columns: string[], row: any[]): CustomerCohort {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) obj[columns[i]] = row[i];\n let curve: number[] = [];\n try { curve = JSON.parse(String(obj.retention_curve || \"[]\")); } catch { curve = []; }\n return {\n id: obj.id,\n cohort_label: obj.cohort_label,\n cohort_size: Number(obj.cohort_size) || 0,\n arpu: Number(obj.arpu) || 0,\n retention_curve: Array.isArray(curve) ? curve.map((n) => Number(n) || 0) : [],\n created_at: obj.created_at,\n };\n}\n\nexport async function listCohorts(): Promise<CustomerCohort[]> {\n const r = await sqlite.execute(\"SELECT id, cohort_label, cohort_size, arpu, retention_curve, created_at FROM customer_cohorts ORDER BY created_at DESC\");\n return (r.rows || []).map((row) => rowToCohort(r.columns, row as any[]));\n}\n\nexport async function createCohort(data: Partial<CustomerCohort>): Promise<CustomerCohort> {\n const id = crypto.randomUUID();\n // QA fix: accept common alias keys (name/label, size/users, ltv/arpu) so the chat tool +\n // UI forms that send shorthand don't silently end up with \"Untitled\" 0/0/0 rows.\n const cohort_label = (data as any).cohort_label || (data as any).name || (data as any).label || \"Untitled\";\n const cohort_size = Number((data as any).cohort_size ?? (data as any).size ?? (data as any).users ?? 0) || 0;\n let arpu = Number((data as any).arpu ?? 0) || 0;\n // If the caller sent an LTV value but no ARPU, back-derive ARPU = LTV / retention_months\n // (default 12) — better than dropping the input.\n if (!arpu && (data as any).ltv) {\n const months = Number((data as any).retention_months ?? 12) || 12;\n arpu = Number((data as any).ltv) / months;\n }\n const curve = JSON.stringify(Array.isArray(data.retention_curve) ? data.retention_curve : []);\n await sqlite.execute(\n \"INSERT INTO customer_cohorts (id, cohort_label, cohort_size, arpu, retention_curve) VALUES (?, ?, ?, ?, ?)\",\n [id, cohort_label, cohort_size, arpu, curve]\n );\n const fetched = await sqlite.execute(\"SELECT id, cohort_label, cohort_size, arpu, retention_curve, created_at FROM customer_cohorts WHERE id = ?\", [id]);\n // QA fix: guard against empty row set so a write that doesn't round-trip returns the\n // intended insert instead of crashing on rows[0].\n if (!fetched.rows?.length) {\n return {\n id,\n cohort_label,\n cohort_size,\n arpu,\n retention_curve: Array.isArray(data.retention_curve) ? data.retention_curve : [],\n created_at: new Date().toISOString(),\n };\n }\n return rowToCohort(fetched.columns, fetched.rows[0] as any[]);\n}\n\nexport async function deleteCohort(id: string): Promise<boolean> {\n await sqlite.execute(\"DELETE FROM customer_cohorts WHERE id = ?\", [id]);\n return true;\n}\n\nexport function computeCohortLTV(cohort: CustomerCohort): number {\n if (!cohort.cohort_size || cohort.cohort_size <= 0 || !cohort.arpu) return 0;\n if (cohort.retention_curve.length === 0) return 0;\n // Geometric sum: LTV = ARPU × Σ (retained_m / cohort_size)\n const retentionRates = cohort.retention_curve.map((r) => r / cohort.cohort_size);\n const survival = retentionRates.reduce((a, b) => a + b, 0);\n return cohort.arpu * survival;\n}\n\nexport async function getCohortSummary(): Promise<{\n cohorts: Array<CustomerCohort & { ltv: number; survival: number }>;\n blended_ltv: number;\n total_customers: number;\n}> {\n const cohorts = await listCohorts();\n const enriched = cohorts.map((c) => {\n const ltv = computeCohortLTV(c);\n const survival = c.retention_curve.length\n ? c.retention_curve.reduce((a, b) => a + b, 0) / (c.cohort_size * c.retention_curve.length)\n : 0;\n return { ...c, ltv, survival };\n });\n const totalCustomers = enriched.reduce((a, c) => a + c.cohort_size, 0);\n const blendedLtv = totalCustomers > 0\n ? enriched.reduce((a, c) => a + c.ltv * c.cohort_size, 0) / totalCustomers\n : 0;\n return { cohorts: enriched, blended_ltv: blendedLtv, total_customers: totalCustomers };\n}\n\n// ─── D3: LEAD ENGAGEMENT VELOCITY ────────────────────────────────────────────\n// Each lead_state_changes row captures a single state transition. Velocity =\n// avg days between MQL and SQL transitions, computed lead-by-lead then averaged.\n\nexport async function recordLeadStateChange(\n lead_email: string,\n to_state: \"cold\" | \"MQL\" | \"SQL\",\n from_state?: \"cold\" | \"MQL\" | \"SQL\",\n): Promise<void> {\n // Skip the insert when the lead is already in this state — avoids noisy duplicate rows\n // every time getLeadScores re-runs and re-classifies the same lead.\n const last = await sqlite.execute(\n \"SELECT to_state FROM lead_state_changes WHERE lead_email = ? ORDER BY changed_at DESC LIMIT 1\",\n [lead_email]\n );\n if (last.rows?.length) {\n const lastState = String((last.rows[0] as any[])[0]);\n if (lastState === to_state) return;\n }\n await sqlite.execute(\n \"INSERT INTO lead_state_changes (lead_email, from_state, to_state) VALUES (?, ?, ?)\",\n [lead_email, from_state ?? null, to_state]\n );\n}\n\nexport async function getEngagementVelocity(): Promise<{\n avg_days_mql_to_sql: number | null;\n avg_days_cold_to_mql: number | null;\n sample_size: number;\n}> {\n const r = await sqlite.execute(\n `SELECT lead_email, to_state, changed_at FROM lead_state_changes ORDER BY lead_email, changed_at ASC`\n );\n if (!r.rows?.length) return { avg_days_mql_to_sql: null, avg_days_cold_to_mql: null, sample_size: 0 };\n\n type Tx = { email: string; state: string; at: number };\n const txs: Tx[] = (r.rows as any[]).map((row) => ({\n email: String(row[0]),\n state: String(row[1]),\n at: new Date(String(row[2])).getTime(),\n }));\n\n const byEmail = new Map<string, Tx[]>();\n for (const t of txs) {\n if (!byEmail.has(t.email)) byEmail.set(t.email, []);\n byEmail.get(t.email)!.push(t);\n }\n\n const mqlToSqlDays: number[] = [];\n const coldToMqlDays: number[] = [];\n for (const [, hist] of byEmail) {\n const mqlTx = hist.find((t) => t.state === \"MQL\");\n const sqlTx = hist.find((t) => t.state === \"SQL\");\n const coldTx = hist.find((t) => t.state === \"cold\");\n if (mqlTx && sqlTx && sqlTx.at > mqlTx.at) {\n mqlToSqlDays.push((sqlTx.at - mqlTx.at) / 86400000);\n }\n if (coldTx && mqlTx && mqlTx.at > coldTx.at) {\n coldToMqlDays.push((mqlTx.at - coldTx.at) / 86400000);\n }\n }\n\n const avg = (arr: number[]) => arr.length ? arr.reduce((a, b) => a + b, 0) / arr.length : null;\n return {\n avg_days_mql_to_sql: avg(mqlToSqlDays),\n avg_days_cold_to_mql: avg(coldToMqlDays),\n sample_size: byEmail.size,\n };\n}\n\n// ─── D5: UTM ATTRIBUTION ─────────────────────────────────────────────────────\n\nexport interface UtmConversion {\n id: number;\n utm_source: string | null;\n utm_medium: string | null;\n utm_campaign: string | null;\n conversion_value: number;\n converted_at: string;\n}\n\nexport async function recordConversion(data: Partial<UtmConversion>): Promise<void> {\n await sqlite.execute(\n \"INSERT INTO utm_conversions (utm_source, utm_medium, utm_campaign, conversion_value) VALUES (?, ?, ?, ?)\",\n [data.utm_source ?? null, data.utm_medium ?? null, data.utm_campaign ?? null, data.conversion_value ?? 0]\n );\n}\n\nexport async function getAttributionSummary(): Promise<{\n channels: Array<{ source: string; conversions: number; value: number; share: number }>;\n total_conversions: number;\n total_value: number;\n window_days: 30;\n}> {\n const r = await sqlite.execute(\n `SELECT COALESCE(utm_source,'direct') AS source, COUNT(*) AS n, COALESCE(SUM(conversion_value), 0) AS v\n FROM utm_conversions\n WHERE converted_at >= datetime('now','-30 days')\n GROUP BY source\n ORDER BY n DESC`\n );\n const rows = (r.rows || []) as any[];\n const channels = rows.map((row) => ({\n source: String(row[0]),\n conversions: Number(row[1]) || 0,\n value: Number(row[2]) || 0,\n share: 0,\n }));\n const totalConv = channels.reduce((a, c) => a + c.conversions, 0);\n const totalVal = channels.reduce((a, c) => a + c.value, 0);\n for (const c of channels) c.share = totalConv > 0 ? c.conversions / totalConv : 0;\n return {\n channels,\n total_conversions: totalConv,\n total_value: totalVal,\n window_days: 30,\n };\n}\n",[],[365,886],"1e29d052-5999-11f1-9cb3-ee650bb23af1","agents.ts","156bf142-56ca-11f1-8a9a-ee650bb23af1",395,["D",1779864959591],"import { sqlite } from \"../db.ts\";\nimport type { Agent } from \"../types.ts\";\n\n// ─── Helpers ─────────────────────────────────────────────────────────────────\n\nfunction rowToAgent(columns: string[], row: any[]): Agent {\n const obj: Record<string, any> = {};\n for (let i = 0; i < columns.length; i++) {\n obj[columns[i]] = row[i];\n }\n return {\n id: obj.id,\n name: obj.name,\n workspace: obj.workspace,\n description: obj.description ?? \"\",\n system_prompt: obj.system_prompt ?? \"\",\n tools: obj.tools ?? \"[]\",\n model: obj.model ?? \"claude-sonnet-4-6\",\n enabled: obj.enabled ?? 1,\n created_at: obj.created_at,\n updated_at: obj.updated_at,\n created_by: obj.created_by ?? \"user\",\n // B2: budget cap + current spend joined from agent_budgets\n monthly_cap: obj.monthly_cap != null ? Number(obj.monthly_cap) : 250,\n current_spend: obj.current_spend != null ? Number(obj.current_spend) : 0,\n // I3: agent role (drives specialized system prompt prefix server-side)\n role: (obj.role as any) ?? \"general\",\n // Campaign fields — written by create_campaign and the Campaigns page.\n channel: obj.channel ?? null,\n goal: obj.goal ?? null,\n audience: obj.audience ?? null,\n status: obj.status ?? \"active\",\n } as Agent;\n}\n\nasync function upsertBudget(agentId: string, monthly_cap: number): Promise<void> {\n // INSERT OR REPLACE would reset current_spend; do upsert-without-clobber instead.\n const existing = await sqlite.execute(\"SELECT agent_id FROM agent_budgets WHERE agent_id = ?\", [agentId]);\n if (existing.rows?.length) {\n await sqlite.execute(\n \"UPDATE agent_budgets SET monthly_cap = ?, updated_at = datetime('now') WHERE agent_id = ?\",\n [monthly_cap, agentId]\n );\n } else {\n await sqlite.execute(\n \"INSERT INTO agent_budgets (agent_id, monthly_cap, current_spend, period_start, updated_at) VALUES (?, ?, 0, datetime('now'), datetime('now'))\",\n [agentId, monthly_cap]\n );\n }\n}\n\n// ─── CRUD functions ───────────────────────────────────────────────────────────\n\nexport async function listAgents(workspace?: string): Promise<Agent[]> {\n let query: string;\n let args: any[];\n\n // B2: LEFT JOIN agent_budgets so monthly_cap/current_spend ride along\n // I3: role column carries the agent's specialized role (general/listener/topic/creator/analyst)\n const base = `SELECT a.id, a.name, a.workspace, a.description, a.system_prompt, a.tools, a.model,\n a.enabled, a.created_at, a.updated_at, a.created_by, a.role,\n a.channel, a.goal, a.audience, a.status,\n b.monthly_cap, b.current_spend\n FROM agents a\n LEFT JOIN agent_budgets b ON b.agent_id = a.id`;\n if (workspace) {\n query = base + \" WHERE a.workspace = ? ORDER BY a.name ASC\";\n args = [workspace];\n } else {\n query = base + \" ORDER BY a.name ASC\";\n args = [];\n }\n\n const result = await sqlite.execute(query, args);\n\n if (!result.rows || result.rows.length === 0) {\n return [];\n }\n\n return result.rows.map((row) => rowToAgent(result.columns, row));\n}\n\nexport async function getAgent(id: string): Promise<Agent | null> {\n const result = await sqlite.execute(\n `SELECT a.id, a.name, a.workspace, a.description, a.system_prompt, a.tools, a.model,\n a.enabled, a.created_at, a.updated_at, a.created_by, a.role,\n a.channel, a.goal, a.audience, a.status,\n b.monthly_cap, b.current_spend\n FROM agents a\n LEFT JOIN agent_budgets b ON b.agent_id = a.id\n WHERE a.id = ?`,\n [id]\n );\n\n if (!result.rows || result.rows.length === 0) {\n return null;\n }\n\n return rowToAgent(result.columns, result.rows[0]);\n}\n\nexport async function createAgent(data: Partial<Agent>): Promise<Agent> {\n const id = crypto.randomUUID();\n const now = new Date().toISOString();\n\n const name = data.name ?? \"Unnamed Agent\";\n const workspace = data.workspace ?? \"marketing\";\n const description = data.description ?? null;\n const system_prompt = data.system_prompt ?? null;\n const tools = (() => {\n if (!data.tools) return \"[]\";\n if (typeof data.tools === \"string\") return data.tools;\n return JSON.stringify(data.tools);\n })();\n const model = (data as any).model ?? \"claude-sonnet-4-6\";\n const created_by = (data as any).created_by ?? \"user\";\n // QA fix: agent role was being silently dropped on create — the role column existed in\n // the schema and chat.ts read it for AGENT_ROLE_PROMPTS injection, but createAgent never\n // wrote it. Listener/Topic/Creator/Analyst roles were always defaulting to 'general'.\n const role = ((data as any).role && String((data as any).role).trim()) || \"general\";\n const channel = (data as any).channel ?? null;\n const goal = (data as any).goal ?? null;\n const audience = (data as any).audience ?? null;\n const status = (data as any).status ?? \"active\";\n\n await sqlite.execute(\n `INSERT INTO agents (id, name, workspace, description, system_prompt, tools, model, enabled, created_at, updated_at, created_by, role, channel, goal, audience, status)\n VALUES (?, ?, ?, ?, ?, ?, ?, 1, ?, ?, ?, ?, ?, ?, ?, ?)`,\n [id, name, workspace, description, system_prompt, tools, model, now, now, created_by, role, channel, goal, audience, status]\n );\n\n // B2: seed budget row (default $250 cap unless caller supplied a monthly_cap)\n const monthly_cap = (data as any).monthly_cap != null && Number.isFinite(Number((data as any).monthly_cap))\n ? Number((data as any).monthly_cap)\n : 250;\n await upsertBudget(id, monthly_cap);\n\n const created = await getAgent(id);\n if (!created) {\n throw new Error(`Failed to retrieve agent after creation (id=${id})`);\n }\n return created;\n}\n\nexport async function updateAgent(\n id: string,\n data: Partial<Agent>\n): Promise<Agent | null> {\n const existing = await getAgent(id);\n if (!existing) {\n return null;\n }\n\n const now = new Date().toISOString();\n\n const name = data.name ?? existing.name;\n const workspace = data.workspace ?? existing.workspace;\n const description = \"description\" in data ? (data.description ?? null) : (existing.description ?? null);\n const system_prompt = \"system_prompt\" in data ? (data.system_prompt ?? null) : (existing.system_prompt ?? null);\n const tools = \"tools\" in data\n ? (typeof data.tools === \"string\" ? data.tools : JSON.stringify(data.tools ?? []))\n : (existing.tools ?? \"[]\");\n const model = (data as any).model ?? (existing as any).model ?? \"claude-sonnet-4-6\";\n const enabled = \"enabled\" in data ? (data as any).enabled : (existing.enabled ?? 1);\n // QA fix: persist role on update too (was being silently dropped).\n const role = \"role\" in (data as any)\n ? (((data as any).role && String((data as any).role).trim()) || \"general\")\n : ((existing as any).role ?? \"general\");\n const channel = \"channel\" in (data as any) ? ((data as any).channel ?? null) : ((existing as any).channel ?? null);\n const goal = \"goal\" in (data as any) ? ((data as any).goal ?? null) : ((existing as any).goal ?? null);\n const audience = \"audience\" in (data as any) ? ((data as any).audience ?? null) : ((existing as any).audience ?? null);\n const status = \"status\" in (data as any) ? ((data as any).status ?? \"active\") : ((existing as any).status ?? \"active\");\n\n await sqlite.execute(\n `UPDATE agents SET name = ?, workspace = ?, description = ?, system_prompt = ?, tools = ?, model = ?, enabled = ?, updated_at = ?, role = ?, channel = ?, goal = ?, audience = ?, status = ? WHERE id = ?`,\n [name, workspace, description, system_prompt, tools, model, enabled, now, role, channel, goal, audience, status, id]\n );\n\n // B2: persist monthly_cap when caller supplies one (omit-to-keep semantics).\n if (\"monthly_cap\" in (data as any)) {\n const cap = Number((data as any).monthly_cap);\n if (Number.isFinite(cap) && cap >= 0) await upsertBudget(id, cap);\n }\n\n return getAgent(id);\n}\n\nexport async function deleteAgent(id: string): Promise<boolean> {\n const existing = await getAgent(id);\n if (!existing) {\n return false;\n }\n\n await sqlite.execute(\"DELETE FROM agents WHERE id = ?\", [id]);\n return true;\n}\n\n// ─── C3: Agent execution log ────────────────────────────────────────────────\nexport async function listAgentRuns(agentId: string, limit = 50): Promise<any[]> {\n const result = await sqlite.execute(\n `SELECT id, agent_id, status, trigger, started_at, finished_at, duration_ms, output, error\n FROM agent_runs WHERE agent_id = ? ORDER BY started_at DESC LIMIT ?`,\n [agentId, limit]\n );\n return (result.rows || []).map((row: any) => {\n const obj: Record<string, any> = {};\n for (let i = 0; i < result.columns.length; i++) obj[result.columns[i]] = row[i];\n return obj;\n });\n}\n\nexport async function startAgentRun(agentId: string, trigger: string, input?: string): Promise<string> {\n const id = crypto.randomUUID();\n await sqlite.execute(\n \"INSERT INTO agent_runs (id, agent_id, status, trigger, started_at, input) VALUES (?, ?, 'running', ?, datetime('now'), ?)\",\n [id, agentId, trigger || 'manual', input ?? null]\n );\n return id;\n}\n\nexport async function finishAgentRun(runId: string, status: 'completed' | 'failed' | 'cancelled', output?: string, error?: string): Promise<void> {\n // Compute duration from started_at recorded server-side to avoid clock drift.\n await sqlite.execute(\n `UPDATE agent_runs\n SET status = ?, finished_at = datetime('now'),\n duration_ms = CAST((julianday('now') - julianday(started_at)) * 86400000 AS INTEGER),\n output = ?, error = ?\n WHERE id = ?`,\n [status, output ?? null, error ?? null, runId]\n );\n}\n",[],[365,894],"51ab9480-5a69-11f1-ad64-ee650bb23af1","actions.ts","51ab957a-5a69-11f1-ad64-ee650bb23af1",429,["D",1779954381335],"// Human-in-the-loop action queue. The AI / agents PREPARE external actions here\n// (send email, publish post, create LinkedIn post, create calendar event). Nothing is\n// sent until a human APPROVES; on approval the real integration call runs (reusing the\n// same handleIntegrationProxy dispatch as the rest of the app), the result is recorded,\n// and any linked task is closed (which propagates to the measure layer).\nimport { sqlite } from \"../db.ts\";\nimport { handleIntegrationProxy, getCredentials } from \"./integrations.ts\";\nimport { updateTask } from \"./tasks.ts\";\n\nexport interface QueuedAction {\n id: number;\n source_type: string;\n source_id: string | null;\n integration: string;\n endpoint: string;\n payload: any;\n preview: string | null;\n title: string | null;\n workspace: string | null;\n status: string; // prepared | approved | executing | done | failed | rejected\n result: any;\n task_id: string | null;\n created_at: string;\n approved_at: string | null;\n executed_at: string | null;\n}\n\nconst COLS = `id, source_type, source_id, integration, endpoint, payload, preview, title,\n workspace, status, result, task_id, created_at, approved_at, executed_at`;\n\nfunction parseMaybe(v: any): any {\n if (v == null) return null;\n try { return JSON.parse(String(v)); } catch { return v; }\n}\n\nfunction rowToAction(cols: string[], row: any[]): QueuedAction {\n const o: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) o[cols[i]] = row[i];\n return {\n id: Number(o.id),\n source_type: String(o.source_type ?? \"chat\"),\n source_id: o.source_id ?? null,\n integration: String(o.integration),\n endpoint: String(o.endpoint),\n payload: parseMaybe(o.payload),\n preview: o.preview ?? null,\n title: o.title ?? null,\n workspace: o.workspace ?? null,\n status: String(o.status ?? \"prepared\"),\n result: parseMaybe(o.result),\n task_id: o.task_id ?? null,\n created_at: String(o.created_at ?? \"\"),\n approved_at: o.approved_at ?? null,\n executed_at: o.executed_at ?? null,\n };\n}\n\nexport async function getAction(id: string | number): Promise<QueuedAction | null> {\n try {\n const r = await sqlite.execute(`SELECT ${COLS} FROM action_queue WHERE id = ?`, [Number(id)]);\n if (!r.rows?.length) return null;\n return rowToAction(r.columns, r.rows[0] as any[]);\n } catch { return null; }\n}\n\nexport async function listActions(status?: string): Promise<QueuedAction[]> {\n try {\n const r = status\n ? await sqlite.execute(`SELECT ${COLS} FROM action_queue WHERE status = ? ORDER BY created_at DESC LIMIT 100`, [status])\n : await sqlite.execute(`SELECT ${COLS} FROM action_queue ORDER BY created_at DESC LIMIT 100`);\n return (r.rows || []).map((row: any) => rowToAction(r.columns, row));\n } catch { return []; }\n}\n\nexport async function enqueueAction(data: {\n integration: string; endpoint: string; payload?: any;\n preview?: string; title?: string; workspace?: string;\n source_type?: string; source_id?: string; task_id?: string;\n}): Promise<QueuedAction | null> {\n if (!data.integration || !data.endpoint) return null;\n await sqlite.execute(\n `INSERT INTO action_queue\n (source_type, source_id, integration, endpoint, payload, preview, title, workspace, task_id, status)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 'prepared')`,\n [\n data.source_type ?? \"chat\", data.source_id ?? null,\n data.integration, data.endpoint,\n JSON.stringify(data.payload ?? {}),\n data.preview ?? null, data.title ?? null,\n data.workspace ?? \"marketing\", data.task_id ?? null,\n ],\n );\n const idRow = await sqlite.execute(\"SELECT last_insert_rowid()\");\n return await getAction(Number((idRow.rows[0] as any[])[0]));\n}\n\nexport async function rejectAction(id: string | number): Promise<QueuedAction | null> {\n await sqlite.execute(\n \"UPDATE action_queue SET status='rejected' WHERE id = ? AND status IN ('prepared','approved')\",\n [Number(id)],\n );\n return await getAction(id);\n}\n\n// Approve → run the real integration call → record result → close linked task.\nexport async function approveAction(id: string | number): Promise<QueuedAction | null> {\n const action = await getAction(id);\n if (!action) return null;\n if (action.status !== \"prepared\" && action.status !== \"approved\") return action;\n\n await sqlite.execute(\n \"UPDATE action_queue SET status='executing', approved_at=datetime('now') WHERE id = ?\",\n [Number(id)],\n );\n\n let ok = false;\n let result: any = null;\n try {\n const creds = await getCredentials(action.integration);\n if (!creds) {\n result = { error: `Integration '${action.integration}' not configured` };\n } else {\n const params = new URLSearchParams({ endpoint: action.endpoint });\n const req = new Request(\n `https://internal/api/integrations/${action.integration}/data?endpoint=${encodeURIComponent(action.endpoint)}`,\n { method: \"POST\", headers: { \"Content-Type\": \"application/json\" }, body: JSON.stringify(action.payload ?? {}) },\n );\n const res = await handleIntegrationProxy(req, action.integration, params, creds);\n result = await res.json().catch(() => ({}));\n ok = res.ok && !(result && result.error);\n }\n } catch (e: any) {\n result = { error: String(e?.message || e) };\n }\n\n await sqlite.execute(\n \"UPDATE action_queue SET status = ?, result = ?, executed_at = datetime('now') WHERE id = ?\",\n [ok ? \"done\" : \"failed\", JSON.stringify(result ?? {}), Number(id)],\n );\n\n // On success, close the linked task — updateTask triggers propagateTaskDone (measure layer).\n if (ok && action.task_id) {\n try { await updateTask(action.task_id, { status: \"done\" } as any); } catch { /* non-critical */ }\n }\n return await getAction(id);\n}\n",[],[365,902],"c7a5c918-5806-11f1-97e7-ee650bb23af1","c7a5c9fe-5806-11f1-97e7-ee650bb23af1","de234ff5-1325-4b93-9eef-d2295e01092a",120,["D",1779692156770],"// YouTube · channels, videos, playlists, search.\n// Docs: https://developers.google.com/youtube/v3\n// Note: mine=true requires OAuth. API key works only for public endpoints.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/youtube/v3\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const hasOAuth = !!creds.access_token;\n const headers: Record<string, string> = {};\n let authParam = \"\";\n\n if (hasOAuth) headers[\"Authorization\"] = `Bearer ${creds.access_token}`;\n else if (creds.api_key) authParam = `key=${encodeURIComponent(creds.api_key)}`;\n\n const jp = (a: string, b: string) => [a, b].filter(Boolean).join(\"&\");\n\n let url: string;\n switch (endpoint) {\n case \"channels\":\n url = hasOAuth\n ? `${BASE}/channels?part=snippet,statistics,brandingSettings,contentDetails&mine=true&${jp(params, authParam)}`\n : `${BASE}/channels?part=snippet,statistics&chart=mostPopular&maxResults=5&${jp(params, authParam)}`;\n break;\n case \"videos\":\n url = `${BASE}/videos?part=snippet,statistics,contentDetails&chart=mostPopular&maxResults=10&${jp(params, authParam)}`;\n break;\n case \"playlists\":\n if (!hasOAuth) return json({ error: \"Playlists requires OAuth. Connect Google in Settings.\" }, 401);\n url = `${BASE}/playlists?part=snippet,contentDetails&mine=true&maxResults=20&${jp(params, authParam)}`;\n break;\n case \"search\":\n url = `${BASE}/search?part=snippet&type=video&maxResults=10&${jp(params, authParam)}`;\n break;\n default:\n url = `${BASE}/${endpoint}?${jp(params, authParam)}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,568],"integrations","58ebcc96-5b19-11f1-8599-ee650bb23af1","c8f8da6c-5806-11f1-9882-ee650bb23af1",467,["D",1780029984924],"// WordPress REST API · posts, pages, media, categories, publish-post.\n// Docs: https://developer.wordpress.org/rest-api/\n// Auth: Basic base64(username:application_password). Requires HTTPS.\nimport { json, safeJson, isSafePublicUrl, IntegrationCredentials } from \"./_shared.ts\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const siteUrl = (creds.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl) return json({ error: \"WordPress site URL required. Add it in Settings.\" }, 400);\n if (!isSafePublicUrl(siteUrl)) return json({ error: \"WordPress site URL must be a public https:// address.\" }, 400);\n // QA fix: refuse to build malformed Basic auth from empty creds (used to silently 401).\n const username = (creds.api_key ?? \"\").trim();\n const appPassword = (creds.api_secret ?? \"\").trim();\n if (!username || !appPassword) {\n return json({ error: \"WordPress requires both a username (api_key) and an application password (api_secret) in Settings.\" }, 400);\n }\n const base = `${siteUrl}/wp-json/wp/v2`;\n const basicAuth = btoa(`${username}:${appPassword}`);\n const headers = { Authorization: `Basic ${basicAuth}`, \"Content-Type\": \"application/json\" };\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"posts\":\n url = `${base}/posts?per_page=20&orderby=date&order=desc&${params}`;\n break;\n case \"pages\":\n url = `${base}/pages?per_page=20&orderby=date&order=desc&${params}`;\n break;\n case \"media\":\n url = `${base}/media?per_page=20&${params}`;\n break;\n case \"categories\":\n url = `${base}/categories?per_page=50&${params}`;\n break;\n case \"publish-post\":\n url = `${base}/posts`;\n method = \"POST\";\n body = await req.text();\n break;\n default:\n url = `${base}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const siteUrl = (creds.extra_config?.site_url ?? \"\").replace(/\\/$/, \"\");\n if (!siteUrl) return { ok: false, message: \"Site URL is required\" };\n if (!isSafePublicUrl(siteUrl)) return { ok: false, message: \"Site URL must be a public https:// address\" };\n const username = (creds.api_key ?? \"\").trim();\n const appPassword = (creds.api_secret ?? \"\").trim();\n if (!username || !appPassword) return { ok: false, message: \"Username + application password required\" };\n const basicAuth = btoa(`${username}:${appPassword}`);\n const res = await fetch(`${siteUrl}/wp-json/wp/v2/users/me`, { headers: { Authorization: `Basic ${basicAuth}` } });\n if (!res.ok) return { ok: false, message: `Auth failed (${res.status})` };\n return { ok: true, message: \"Credentials verified\" };\n}\n",[],[365,917,576],"6d858758-5b48-11f1-94b4-ee650bb23af1","6d85882a-5b48-11f1-94b4-ee650bb23af1",499,["D",1780050205835],"// Microsoft Teams (Microsoft Graph) · teams, channels, channel messages, chats, chat messages.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/teams-api-overview\n// READ: teams (joinedTeams), channels, messages, chats, chat-messages\n// WRITE: send-channel, send-chat\n// Shares the central `outlook` Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n // Id-bearing reads + writes accept identifiers from the POST body (the panel form) or the query\n // string. Read the body once and pick from either (query-first).\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"teams\":\n url = `${BASE}/me/joinedTeams`;\n break;\n case \"channels\": {\n const teamId = pick(\"teamId\");\n if (!teamId) return json({ error: \"teamId required\" }, 400);\n url = `${BASE}/teams/${encodeURIComponent(teamId)}/channels`;\n break;\n }\n case \"messages\": {\n const teamId = pick(\"teamId\");\n const channelId = pick(\"channelId\");\n if (!teamId || !channelId) return json({ error: \"teamId and channelId required\" }, 400);\n url = `${BASE}/teams/${encodeURIComponent(teamId)}/channels/${encodeURIComponent(channelId)}/messages?$top=30`;\n break;\n }\n case \"chats\":\n url = `${BASE}/me/chats?$top=30`;\n break;\n case \"chat-messages\": {\n const chatId = pick(\"chatId\");\n if (!chatId) return json({ error: \"chatId required\" }, 400);\n url = `${BASE}/me/chats/${encodeURIComponent(chatId)}/messages?$top=30`;\n break;\n }\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"send-channel\": {\n const teamId = pick(\"teamId\");\n const channelId = pick(\"channelId\");\n if (!teamId || !channelId) return json({ error: \"teamId and channelId required\" }, 400);\n if (!bodyParams.content) return json({ error: \"content required\" }, 400);\n const res = await fetch(`${BASE}/teams/${encodeURIComponent(teamId)}/channels/${encodeURIComponent(channelId)}/messages`, {\n method: \"POST\", headers,\n body: JSON.stringify({ body: { contentType: \"html\", content: String(bodyParams.content) } }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"send-chat\": {\n const chatId = pick(\"chatId\");\n if (!chatId) return json({ error: \"chatId required\" }, 400);\n if (!bodyParams.content) return json({ error: \"content required\" }, 400);\n const res = await fetch(`${BASE}/me/chats/${encodeURIComponent(chatId)}/messages`, {\n method: \"POST\", headers,\n body: JSON.stringify({ body: { content: String(bodyParams.content) } }),\n });\n return json(await safeJson(res), res.status);\n }\n\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(`${BASE}/me/joinedTeams`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,584],"c64033d8-5806-11f1-8516-ee650bb23af1","c64034aa-5806-11f1-8516-ee650bb23af1",117,["D",1779692154414],"// Google Sheets · list spreadsheets, read values, create/append/update.\n// Docs: https://developers.google.com/sheets/api/reference/rest\n// READ: list (via Drive query), values, metadata, batchGet\n// WRITE: create, appendRow, updateCells\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst SHEETS_BASE = \"https://sheets.googleapis.com/v4/spreadsheets\";\nconst DRIVE_BASE = \"https://www.googleapis.com/drive/v3/files\";\nconst FULL_FILES = \"files(id,name,mimeType,modifiedTime,createdTime,size,webViewLink,iconLink,owners(displayName,emailAddress),lastModifyingUser(displayName,emailAddress),shared,starred),nextPageToken\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"list\": {\n const driveSp = new URLSearchParams({\n q: \"mimeType='application/vnd.google-apps.spreadsheet' and trashed=false\",\n pageSize: sp.get(\"pageSize\") || \"50\",\n orderBy: \"modifiedTime desc\",\n fields: FULL_FILES,\n });\n url = `${DRIVE_BASE}?${driveSp.toString()}`;\n break;\n }\n case \"values\": {\n const id = sp.get(\"spreadsheetId\");\n const range = sp.get(\"range\") || \"A1:Z100\";\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n const opt = sp.get(\"valueRenderOption\") || \"FORMATTED_VALUE\";\n url = `${SHEETS_BASE}/${id}/values/${encodeURIComponent(range)}?valueRenderOption=${opt}&dateTimeRenderOption=FORMATTED_STRING`;\n break;\n }\n case \"metadata\": {\n const id = sp.get(\"spreadsheetId\");\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n url = `${SHEETS_BASE}/${id}?includeGridData=false`;\n break;\n }\n case \"batchGet\": {\n const id = sp.get(\"spreadsheetId\");\n if (!id) return json({ error: \"spreadsheetId param required\" }, 400);\n const ranges = sp.getAll(\"ranges\");\n const rangesQ = ranges.length ? ranges.map(r => `ranges=${encodeURIComponent(r)}`).join(\"&\") : \"ranges=A1:Z1000\";\n url = `${SHEETS_BASE}/${id}/values:batchGet?${rangesQ}&valueRenderOption=FORMATTED_VALUE`;\n break;\n }\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"create\": {\n const body = await req.json().catch(() => ({}));\n if (!body.title) return json({ error: \"title required\" }, 400);\n const res = await fetch(SHEETS_BASE, {\n method: \"POST\", headers,\n body: JSON.stringify({ properties: { title: body.title } }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"appendRow\": {\n const body = await req.json().catch(() => ({}));\n if (!body.spreadsheetId) return json({ error: \"spreadsheetId required\" }, 400);\n const range = body.range || \"Sheet1!A:Z\";\n const values = Array.isArray(body.values?.[0]) ? body.values : [body.values || []];\n const res = await fetch(\n `${SHEETS_BASE}/${body.spreadsheetId}/values/${encodeURIComponent(range)}:append?valueInputOption=USER_ENTERED&insertDataOption=INSERT_ROWS`,\n { method: \"POST\", headers, body: JSON.stringify({ values }) }\n );\n return json(await safeJson(res), res.status);\n }\n case \"updateCells\": {\n const body = await req.json().catch(() => ({}));\n if (!body.spreadsheetId || !body.range) return json({ error: \"spreadsheetId + range required\" }, 400);\n const res = await fetch(\n `${SHEETS_BASE}/${body.spreadsheetId}/values/${encodeURIComponent(body.range)}?valueInputOption=USER_ENTERED`,\n { method: \"PUT\", headers, body: JSON.stringify({ values: body.values }) }\n );\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${SHEETS_BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,592],"c96adf54-5806-11f1-8bf1-ee650bb23af1","c96ae012-5806-11f1-8bf1-ee650bb23af1",124,["D",1779692159729],"// Product Hunt · GraphQL API for today's posts, topics.\n// Docs: https://api.producthunt.com/v2/docs\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst URL = \"https://api.producthunt.com/v2/api/graphql\";\n\nexport async function handle(_req: Request, endpoint: string, searchParams: URLSearchParams, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.api_key}`, \"Content-Type\": \"application/json\" };\n\n let query: string;\n switch (endpoint) {\n case \"posts\":\n query = `{ posts(first: 20, order: RANKING) { edges { node { id name tagline votesCount commentsCount url createdAt thumbnail { url } topics { edges { node { name } } } } } } }`;\n break;\n case \"topics\":\n query = `{ topics(first: 30) { edges { node { id name slug followersCount } } } }`;\n break;\n default:\n query = searchParams.get(\"query\") ?? `{ posts(first: 10, order: RANKING) { edges { node { id name tagline votesCount } } } }`;\n }\n\n const res = await fetch(URL, { method: \"POST\", headers, body: JSON.stringify({ query }) });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(URL, {\n method: \"POST\",\n headers: { Authorization: `Bearer ${creds.api_key}`, \"Content-Type\": \"application/json\" },\n body: JSON.stringify({ query: \"{ viewer { id } }\" }),\n });\n if (!res.ok) return { ok: false, message: \"Invalid API key\" };\n const d = await safeJson(res);\n if (d.errors?.some((e: any) => /auth|unauthorized/i.test(e.message ?? \"\"))) return { ok: false, message: \"Invalid API key\" };\n return { ok: true, message: \"API key verified\" };\n}\n",[],[365,917,600],"255707c2-5d69-11f1-8f15-1607ee4eb77e","c88a5420-5806-11f1-b58f-ee650bb23af1",617,["D",1780284160447],"// Outlook (Microsoft Graph) · mail, calendar, contacts.\n// Docs: https://learn.microsoft.com/en-us/graph/api/overview\n// READ: inbox/messages/calendar/contacts\n// WRITE: send\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"inbox\":\n case \"messages\":\n url = `${BASE}/messages?$top=20&$orderby=receivedDateTime desc&${params}`;\n break;\n case \"send\":\n url = `${BASE}/sendMail`;\n method = \"POST\";\n body = await req.text();\n break;\n case \"calendar\":\n url = `${BASE}/calendar/events?$top=20&$orderby=start/dateTime asc&${params}`;\n break;\n case \"contacts\":\n url = `${BASE}/contacts?$top=20&${params}`;\n break;\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,608],"2502124e-5d69-11f1-ba26-1607ee4eb77e","6c4c2216-5b48-11f1-bc6f-ee650bb23af1",616,["D",1780284159879],"// OneDrive (Microsoft Graph) · browse, search, and create files & folders.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/onedrive\n// READ: recent / files / search WRITE: createFolder\n// Shares the central `outlook` Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me/drive\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n // Form-based actions (search, createFolder) POST a JSON body from the panel; direct/AI calls\n // may pass query params. Read the body once and accept either source (query-first).\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n case \"recent\":\n url = `${BASE}/recent`;\n break;\n case \"files\":\n url = `${BASE}/root/children?$top=100&$orderby=lastModifiedDateTime desc`;\n break;\n case \"search\": {\n const q = String(pick(\"q\") || pick(\"query\") || \"\").replace(/'/g, \"''\");\n url = `${BASE}/root/search(q='${encodeURIComponent(q)}')`;\n break;\n }\n case \"createFolder\": {\n const name = pick(\"name\");\n if (!name) return json({ error: \"name required\" }, 400);\n const res = await fetch(`${BASE}/root/children`, {\n method: \"POST\", headers,\n body: JSON.stringify({ name, folder: {}, \"@microsoft.graph.conflictBehavior\": \"rename\" }),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,616],"24a2afc0-5d69-11f1-8218-1607ee4eb77e","c8228bd8-5806-11f1-a99e-ee650bb23af1",615,["D",1780284159258],16,"// LinkedIn v2 API · profile, posts, connections.\n// Docs: https://learn.microsoft.com/en-us/linkedin/marketing/integrations/community-management/shares/ugc-post-api\n// Most endpoints require Marketing Developer Platform approval.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://api.linkedin.com/v2\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n // QA fix: guard missing access_token instead of sending 'Bearer undefined'\n if (!creds.access_token) {\n return json({ error: \"LinkedIn access token required. Connect LinkedIn under Settings → LinkedIn.\" }, 401);\n }\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const authorId = (creds.extra_config?.author_id ?? \"\").toString().trim();\n\n let url: string;\n switch (endpoint) {\n case \"profile\":\n url = `${BASE}/me?projection=(id,firstName,lastName,profilePicture)`;\n break;\n case \"posts\":\n // QA fix: posts requires an author_id — without it LinkedIn returns a confusing 400.\n if (!authorId) return json({ error: \"LinkedIn 'posts' requires author_id (your urn:li:person:... or urn:li:organization:...). Add it under Settings → LinkedIn.\" }, 400);\n url = `${BASE}/ugcPosts?q=authors&authors=List(${authorId})&${params}`;\n break;\n case \"analytics\":\n url = `${BASE}/organizationalEntityShareStatistics?${params}`;\n break;\n case \"connections\":\n url = `${BASE}/connections?q=viewer&${params}`;\n break;\n // ─── WRITE ────────────────────────────────────────────────────────────────\n case \"create-post\": {\n if (!authorId) return json({ error: \"LinkedIn 'create-post' requires author_id (urn:li:person:... or urn:li:organization:...). Add it under Settings → LinkedIn.\" }, 400);\n const body = await _req.json().catch(() => ({}));\n const text = String(body.text ?? body.content ?? \"\").trim();\n if (!text) return json({ error: \"text required\" }, 400);\n const payload = {\n author: authorId,\n lifecycleState: \"PUBLISHED\",\n specificContent: {\n \"com.linkedin.ugc.ShareContent\": {\n shareCommentary: { text },\n shareMediaCategory: \"NONE\",\n },\n },\n visibility: { \"com.linkedin.ugc.MemberNetworkVisibility\": \"PUBLIC\" },\n };\n const res = await fetch(`${BASE}/ugcPosts`, {\n method: \"POST\",\n headers: { ...headers, \"X-Restli-Protocol-Version\": \"2.0.0\" },\n body: JSON.stringify(payload),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n if (!creds.access_token) return { ok: false, message: \"Access token required\" };\n const res = await fetch(`${BASE}/me`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n\nexport async function exchangeLinkedInCode(code: string, redirectUri: string, clientId: string, clientSecret: string): Promise<string> {\n const res = await fetch(\"https://www.linkedin.com/oauth/v2/accessToken\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({\n grant_type: \"authorization_code\",\n code,\n redirect_uri: redirectUri,\n client_id: clientId,\n client_secret: clientSecret,\n }),\n });\n const data = await res.json() as any;\n if (!res.ok || !data.access_token) {\n throw new Error(data.error_description || data.error || \"LinkedIn token exchange failed\");\n }\n return data.access_token;\n}\n",[],[365,917,624],"a44abc86-5968-11f1-a954-ee650bb23af1","c4da5e06-5806-11f1-b423-ee650bb23af1",311,["D",1779844139288],"// Google Search Console · query, page, and sitemap data.\n// Docs: https://developers.google.com/webmaster-tools/v1/api_reference_index\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://searchconsole.googleapis.com/webmasters/v3/sites\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n // QA fix: guard required fields up front instead of constructing a malformed URL.\n const rawSiteUrl = (creds.extra_config?.site_url ?? \"\").toString().trim();\n if (!rawSiteUrl) {\n return json({ error: \"Search Console site URL required. Add 'Site URL' under Settings → GSC (e.g. https://example.com or sc-domain:example.com).\" }, 400);\n }\n if (!creds.access_token) {\n return json({ error: \"Search Console needs Google OAuth. Click 'Connect with Google' in Settings.\" }, 401);\n }\n const siteUrl = encodeURIComponent(rawSiteUrl);\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n let method = \"GET\";\n let body: string | undefined;\n\n switch (endpoint) {\n case \"search-analytics\":\n case \"searchanalytics\": {\n url = `${BASE}/${siteUrl}/searchAnalytics/query`;\n method = \"POST\";\n const saDays = parseInt(sp.get(\"days\") || \"90\", 10);\n const saStartDate = new Date(Date.now() - saDays * 86400000).toISOString().slice(0, 10);\n body = (await req.text()) || JSON.stringify({\n startDate: saStartDate,\n endDate: new Date().toISOString().slice(0, 10),\n dimensions: [\"query\"],\n rowLimit: 50,\n });\n break;\n }\n case \"pages\": {\n url = `${BASE}/${siteUrl}/searchAnalytics/query`;\n method = \"POST\";\n const pgDays = parseInt(sp.get(\"days\") || \"90\", 10);\n const pgStartDate = new Date(Date.now() - pgDays * 86400000).toISOString().slice(0, 10);\n body = JSON.stringify({\n startDate: pgStartDate,\n endDate: new Date().toISOString().slice(0, 10),\n dimensions: [\"page\"],\n rowLimit: 25,\n });\n break;\n }\n case \"sitemaps\":\n url = `${BASE}/${siteUrl}/sitemaps`;\n break;\n default:\n url = `${BASE}/${siteUrl}/${endpoint}`;\n }\n\n const res = await fetch(url, { method, headers, body });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,632],"c54abd2c-5806-11f1-8bf1-ee650bb23af1","c54abdfe-5806-11f1-8bf1-ee650bb23af1",115,["D",1779692152815],"// Gmail · inbox + drafts + send + label management.\n// Docs: https://developers.google.com/gmail/api/reference/rest\n// READ: inbox/unread/sent/starred/labels/profile/threads/message?id=X\n// WRITE: send, createDraft, modifyLabels\n// List endpoints auto-hydrate the first 20 messages with sender/subject/date/labelIds/snippet/size.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://gmail.googleapis.com/gmail/v1/users/me\";\nconst STD_HEADERS = [\"From\", \"To\", \"Cc\", \"Bcc\", \"Subject\", \"Date\", \"Reply-To\", \"Message-ID\"];\n\n// RFC 2822 → base64url for Gmail's `raw` param.\nfunction encodeRaw(headers: string[], body: string): string {\n return btoa(unescape(encodeURIComponent([...headers, \"Content-Type: text/plain; charset=UTF-8\", \"\", body].join(\"\\r\\n\"))))\n .replace(/\\+/g, \"-\").replace(/\\//g, \"_\").replace(/=+$/, \"\");\n}\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n const HYDRATE_LIMIT = parseInt(sp.get(\"hydrate\") || \"20\", 10);\n sp.delete(\"hydrate\");\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"inbox\":\n case \"messages\":\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n if (!sp.has(\"q\")) sp.set(\"q\", \"in:inbox\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"threads\":\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/threads?${sp.toString()}`;\n break;\n case \"labels\":\n url = `${BASE}/labels`;\n break;\n case \"profile\":\n url = `${BASE}/profile`;\n break;\n case \"unread\":\n sp.set(\"q\", \"is:unread in:inbox\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"sent\":\n sp.set(\"q\", \"in:sent\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n case \"starred\":\n sp.set(\"q\", \"is:starred\");\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n url = `${BASE}/messages?${sp.toString()}`;\n break;\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"send\": {\n const body = await req.json().catch(() => ({}));\n if (!body.to || !body.subject) return json({ error: \"to + subject required\" }, 400);\n const lines = [`To: ${body.to}`];\n if (body.cc) lines.push(`Cc: ${body.cc}`);\n if (body.bcc) lines.push(`Bcc: ${body.bcc}`);\n lines.push(`Subject: ${body.subject}`);\n const raw = encodeRaw(lines, body.body || \"\");\n const res = await fetch(`${BASE}/messages/send`, { method: \"POST\", headers, body: JSON.stringify({ raw }) });\n return json(await safeJson(res), res.status);\n }\n case \"createDraft\": {\n const body = await req.json().catch(() => ({}));\n const lines: string[] = [];\n if (body.to) lines.push(`To: ${body.to}`);\n if (body.cc) lines.push(`Cc: ${body.cc}`);\n if (body.subject) lines.push(`Subject: ${body.subject}`);\n const raw = encodeRaw(lines, body.body || \"\");\n const res = await fetch(`${BASE}/drafts`, { method: \"POST\", headers, body: JSON.stringify({ message: { raw } }) });\n return json(await safeJson(res), res.status);\n }\n case \"modifyLabels\": {\n const body = await req.json().catch(() => ({}));\n if (!body.id) return json({ error: \"id required\" }, 400);\n const res = await fetch(`${BASE}/messages/${body.id}/modify`, {\n method: \"POST\", headers,\n body: JSON.stringify({ addLabelIds: body.addLabelIds || [], removeLabelIds: body.removeLabelIds || [] }),\n });\n return json(await safeJson(res), res.status);\n }\n\n default:\n if (endpoint === \"message\" && sp.get(\"id\")) {\n url = `${BASE}/messages/${sp.get(\"id\")}?format=full`;\n } else {\n url = `${BASE}/${endpoint}?${params}`;\n }\n }\n\n const res = await fetch(url, { headers });\n let data = await safeJson(res);\n\n // Hydrate list endpoints with full standard headers + labels + snippet.\n const listEndpoints = [\"inbox\", \"messages\", \"unread\", \"sent\", \"starred\"];\n if (listEndpoints.includes(endpoint) && Array.isArray(data?.messages)) {\n const headerQs = STD_HEADERS.map(h => `metadataHeaders=${encodeURIComponent(h)}`).join(\"&\");\n const limited = data.messages.slice(0, HYDRATE_LIMIT);\n const hydrated = await Promise.all(limited.map(async (m: any) => {\n try {\n const r = await fetch(`${BASE}/messages/${m.id}?format=metadata&${headerQs}`, { headers });\n const d = await r.json();\n const hdrs: any[] = d.payload?.headers || [];\n const get = (n: string) => hdrs.find((h) => h.name?.toLowerCase() === n.toLowerCase())?.value ?? \"\";\n return {\n id: m.id, threadId: m.threadId,\n from: get(\"From\"), to: get(\"To\"), cc: get(\"Cc\"),\n subject: get(\"Subject\"), date: get(\"Date\"),\n replyTo: get(\"Reply-To\"), messageId: get(\"Message-ID\"),\n snippet: d.snippet || \"\",\n labelIds: d.labelIds || [],\n sizeEstimate: d.sizeEstimate,\n internalDate: d.internalDate,\n unread: (d.labelIds || []).includes(\"UNREAD\"),\n starred: (d.labelIds || []).includes(\"STARRED\"),\n };\n } catch (e) { return { id: m.id, threadId: m.threadId, error: String(e) }; }\n }));\n data = { ...data, _hydrated: hydrated };\n }\n return json(data, res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(`${BASE}/profile`, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: `Gmail unreachable (${res.status})` };\n const p = await safeJson(res);\n return { ok: true, message: `Gmail OK · ${p.emailAddress} · ${p.messagesTotal} messages` };\n}\n",[],[365,917,640],"a3c5be78-5968-11f1-bbfa-ee650bb23af1","c4608676-5806-11f1-9882-ee650bb23af1",310,["D",1779844138416],"// Google Analytics 4 · property-level reports (sessions, users, events, funnels, KPIs).\n// Docs: https://developers.google.com/analytics/devguides/reporting/data/v1/rest\n// Auth: Bearer access_token (shared with central Google integration).\n// All reporting endpoints use POST.\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst V1 = `https://analyticsdata.googleapis.com/v1beta/properties`;\nconst V1ALPHA = `https://analyticsdata.googleapis.com/v1alpha/properties`;\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const propertyId = (creds.extra_config?.property_id ?? \"\").toString().trim();\n // QA fix: refuse to build malformed URL like /properties//runReport — used to 400 silently.\n if (!propertyId) {\n return json({ error: \"GA4 property ID required. Add 'Property ID' under Settings → GA4 (e.g. 123456789).\" }, 400);\n }\n if (!creds.access_token) {\n return json({ error: \"GA4 needs Google OAuth. Click 'Connect with Google' in Settings.\" }, 401);\n }\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string, body: string;\n\n switch (endpoint) {\n case \"realtime\":\n url = `${V1}/${propertyId}:runRealtimeReport`;\n body = (await req.text()) || JSON.stringify({\n dimensions: [{ name: \"country\" }],\n metrics: [{ name: \"activeUsers\" }],\n minuteRanges: [{ name: \"last30Minutes\", startMinutesAgo: 29, endMinutesAgo: 0 }],\n });\n break;\n case \"report\": {\n url = `${V1}/${propertyId}:runReport`;\n const reportDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const reportStartDate = `${reportDays}daysAgo`;\n body = (await req.text()) || JSON.stringify({\n dimensions: [{ name: \"sessionDefaultChannelGroup\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"eventCount\" }],\n dateRanges: [{ startDate: reportStartDate, endDate: \"today\" }],\n });\n break;\n }\n case \"user-acquisition\": {\n url = `${V1}/${propertyId}:runReport`;\n const uaDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const uaStartDate = `${uaDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"sessionDefaultChannelGroup\" }],\n metrics: [{ name: \"sessions\" }, { name: \"newUsers\" }, { name: \"engagementRate\" }],\n dateRanges: [{ startDate: uaStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"sessions\" }, desc: true }],\n limit: \"20\",\n });\n break;\n }\n case \"top-events\": {\n url = `${V1}/${propertyId}:runReport`;\n const teDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const teStartDate = `${teDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"eventName\" }],\n metrics: [{ name: \"eventCount\" }, { name: \"totalUsers\" }],\n dateRanges: [{ startDate: teStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"eventCount\" }, desc: true }],\n limit: \"20\",\n });\n break;\n }\n case \"kpi\": {\n url = `${V1}/${propertyId}:runReport`;\n const kpiDays = sp.get(\"days\") || \"30\";\n body = JSON.stringify({\n metrics: [\n { name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"newUsers\" },\n { name: \"engagementRate\" }, { name: \"screenPageViews\" }, { name: \"bounceRate\" },\n ],\n dateRanges: [\n { name: \"current\", startDate: `${kpiDays}daysAgo`, endDate: \"today\" },\n { name: \"previous\", startDate: `${Number(kpiDays) * 2}daysAgo`, endDate: `${Number(kpiDays) + 1}daysAgo` },\n ],\n });\n break;\n }\n case \"traffic-over-time\": {\n url = `${V1}/${propertyId}:runReport`;\n const totDays = sp.get(\"days\") || \"30\";\n body = JSON.stringify({\n dimensions: [{ name: \"date\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }],\n dateRanges: [{ startDate: `${totDays}daysAgo`, endDate: \"today\" }],\n orderBys: [{ dimension: { dimensionName: \"date\" }, desc: false }],\n limit: \"90\",\n });\n break;\n }\n case \"device-breakdown\": {\n url = `${V1}/${propertyId}:runReport`;\n const dbDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const dbStartDate = `${dbDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"deviceCategory\" }],\n metrics: [{ name: \"sessions\" }, { name: \"activeUsers\" }, { name: \"bounceRate\" }],\n dateRanges: [{ startDate: dbStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"sessions\" }, desc: true }],\n });\n break;\n }\n case \"page-views\": {\n url = `${V1}/${propertyId}:runReport`;\n const pvDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const pvStartDate = `${pvDays}daysAgo`;\n body = JSON.stringify({\n dimensions: [{ name: \"pagePath\" }],\n metrics: [{ name: \"screenPageViews\" }, { name: \"sessions\" }, { name: \"bounceRate\" }],\n dateRanges: [{ startDate: pvStartDate, endDate: \"today\" }],\n orderBys: [{ metric: { metricName: \"screenPageViews\" }, desc: true }],\n limit: \"25\",\n });\n break;\n }\n case \"funnel\": {\n url = `${V1ALPHA}/${propertyId}:runFunnelReport`;\n const fnDays = parseInt(sp.get(\"days\") || \"30\", 10);\n const fnStartDate = `${fnDays}daysAgo`;\n body = (await req.text()) || JSON.stringify({\n dateRanges: [{ startDate: fnStartDate, endDate: \"today\" }],\n funnel: {\n steps: [\n { name: \"Session Start\", filterExpression: { funnelEventFilter: { eventName: \"session_start\" } } },\n { name: \"Sign Up\", filterExpression: { funnelEventFilter: { eventName: \"sign_up\" } } },\n ],\n },\n });\n break;\n }\n default:\n url = `${V1}/${propertyId}/${endpoint}`;\n body = await req.text();\n }\n\n const res = await fetch(url, { method: \"POST\", headers, body });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,648],"6ddf4bd0-5b48-11f1-bbb6-ee650bb23af1","6ddf4c84-5b48-11f1-bbb6-ee650bb23af1",500,["D",1780050206416],"// Microsoft Excel (Microsoft Graph) · browse workbooks, list worksheets, read cell ranges.\n// Docs: https://learn.microsoft.com/en-us/graph/api/resources/excel\n// READ: workbooks (list .xlsx) / worksheets (by file id) / range (used range of a sheet)\n// Shares the central Outlook Graph token (see MICROSOFT_SERVICES in _shared.ts).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://graph.microsoft.com/v1.0/me/drive\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n // worksheets/range carry their identifiers either in the query string (direct/AI calls)\n // or in a POST body (the panel renders a form for these). Accept both, query-first.\n let bodyParams: Record<string, any> = {};\n if (req.method === \"POST\") bodyParams = await req.json().catch(() => ({}));\n const pick = (k: string) => sp.get(k) || bodyParams[k] || \"\";\n\n let url: string;\n switch (endpoint) {\n case \"workbooks\": {\n // List Excel workbooks in the user's OneDrive (search by extension).\n url = `${BASE}/root/search(q='.xlsx')?$top=50&$select=id,name,webUrl,lastModifiedDateTime,size`;\n break;\n }\n case \"worksheets\": {\n const id = pick(\"id\");\n if (!id) return json({ error: \"id required (a workbook file id from Workbooks)\" }, 400);\n url = `${BASE}/items/${encodeURIComponent(id)}/workbook/worksheets`;\n break;\n }\n case \"range\": {\n const id = pick(\"id\");\n const sheet = pick(\"sheet\");\n if (!id) return json({ error: \"id required (a workbook file id from Workbooks)\" }, 400);\n if (!sheet) return json({ error: \"sheet required (a worksheet name from Worksheets)\" }, 400);\n url = `${BASE}/items/${encodeURIComponent(id)}/workbook/worksheets/${encodeURIComponent(sheet)}/usedRange?$select=address,values`;\n break;\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const res = await fetch(BASE, { headers: { Authorization: `Bearer ${creds.access_token}` } });\n if (!res.ok) return { ok: false, message: \"Invalid access token\" };\n return { ok: true, message: \"Access token verified\" };\n}\n",[],[365,917,656],"22d42ed0-5d69-11f1-ad33-1607ee4eb77e","c5b2d3da-5806-11f1-8f42-ee650bb23af1",614,["D",1780284156235],"// Google Drive · list/search/create/delete files & folders.\n// Docs: https://developers.google.com/drive/api/reference/rest/v3\n// READ: recent/shared/starred/trash/search/about\n// WRITE: createFolder, deleteFile\n// Returns full file metadata (owners, permissions, sharing, capabilities, version, md5Checksum).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/drive/v3\";\nconst FULL = \"files(id,name,mimeType,description,starred,trashed,parents,modifiedTime,createdTime,viewedByMeTime,size,webViewLink,webContentLink,iconLink,thumbnailLink,owners(displayName,emailAddress,photoLink),lastModifyingUser(displayName,emailAddress,photoLink),sharingUser,shared,permissions(role,type,emailAddress),capabilities,version,md5Checksum,quotaBytesUsed),nextPageToken,incompleteSearch\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n // ─── READ ───────────────────────────────────────────────────────────────\n case \"files\":\n case \"recent\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n if (!sp.has(\"orderBy\")) sp.set(\"orderBy\", \"modifiedTime desc\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"shared\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"sharedWithMe = true\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"starred\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"starred = true and trashed = false\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"trash\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n sp.set(\"q\", \"trashed = true\");\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"search\":\n if (!sp.has(\"pageSize\")) sp.set(\"pageSize\", \"50\");\n if (!sp.has(\"q\") && sp.has(\"query\")) sp.set(\"q\", `name contains '${sp.get(\"query\")?.replace(/'/g, \"\\\\'\")}'`);\n sp.set(\"fields\", FULL);\n url = `${BASE}/files?${sp.toString()}`;\n break;\n case \"about\":\n url = `${BASE}/about?fields=*`;\n break;\n\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"createFolder\": {\n const body = await req.json().catch(() => ({}));\n if (!body.name) return json({ error: \"name required\" }, 400);\n const res = await fetch(`${BASE}/files`, {\n method: \"POST\", headers,\n body: JSON.stringify({ name: body.name, mimeType: \"application/vnd.google-apps.folder\", parents: body.parents || undefined }),\n });\n return json(await safeJson(res), res.status);\n }\n case \"deleteFile\": {\n const body = await req.json().catch(() => ({}));\n if (!body.id) return json({ error: \"id required\" }, 400);\n const res = await fetch(`${BASE}/files/${body.id}`, { method: \"DELETE\", headers });\n if (res.status === 204) return json({ ok: true, deleted: body.id });\n return json(await safeJson(res), res.status);\n }\n\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,664],"2275de52-5d69-11f1-928d-1607ee4eb77e","c6da0d96-5806-11f1-8602-ee650bb23af1",613,["D",1780284155604],"// Google Docs · list documents, create new, append text.\n// Docs: https://developers.google.com/docs/api/reference/rest\n// READ: list (via Drive), document?id=X\n// WRITE: create, appendText\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst DOCS_BASE = \"https://docs.googleapis.com/v1/documents\";\nconst DRIVE_BASE = \"https://www.googleapis.com/drive/v3/files\";\nconst FULL_FILES = \"files(id,name,mimeType,modifiedTime,createdTime,size,webViewLink,iconLink,owners(displayName,emailAddress),lastModifyingUser(displayName,emailAddress),shared,starred,description),nextPageToken\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"list\": {\n const driveSp = new URLSearchParams({\n q: \"mimeType='application/vnd.google-apps.document' and trashed=false\",\n pageSize: sp.get(\"pageSize\") || \"50\",\n orderBy: \"modifiedTime desc\",\n fields: FULL_FILES,\n });\n url = `${DRIVE_BASE}?${driveSp.toString()}`;\n break;\n }\n case \"document\": {\n const id = sp.get(\"documentId\");\n if (!id) return json({ error: \"documentId param required\" }, 400);\n url = `${DOCS_BASE}/${id}`;\n break;\n }\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"create\": {\n const body = await req.json().catch(() => ({}));\n if (!body.title) return json({ error: \"title required\" }, 400);\n const createRes = await fetch(DOCS_BASE, {\n method: \"POST\", headers, body: JSON.stringify({ title: body.title }),\n });\n const doc = await safeJson(createRes);\n if (createRes.ok && body.content && doc.documentId) {\n await fetch(`${DOCS_BASE}/${doc.documentId}:batchUpdate`, {\n method: \"POST\", headers,\n body: JSON.stringify({ requests: [{ insertText: { location: { index: 1 }, text: body.content } }] }),\n });\n }\n return json(doc, createRes.status);\n }\n case \"appendText\": {\n const body = await req.json().catch(() => ({}));\n if (!body.documentId || !body.text) return json({ error: \"documentId + text required\" }, 400);\n const docRes = await fetch(`${DOCS_BASE}/${body.documentId}?fields=body.content`, { headers });\n const docData = await safeJson(docRes);\n const endIndex = docData?.body?.content?.slice(-1)?.[0]?.endIndex || 1;\n const res = await fetch(`${DOCS_BASE}/${body.documentId}:batchUpdate`, {\n method: \"POST\", headers,\n body: JSON.stringify({ requests: [{ insertText: { location: { index: endIndex - 1 }, text: body.text } }] }),\n });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${DOCS_BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,672],"c73dcd86-5806-11f1-ada3-ee650bb23af1","c73dce58-5806-11f1-ada3-ee650bb23af1",119,["D",1779692156085],"// Google Calendar · list/create/delete events.\n// Docs: https://developers.google.com/calendar/api/v3/reference\n// READ: events/upcoming/today/week/calendars/settings\n// WRITE: createEvent, deleteEvent\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://www.googleapis.com/calendar/v3\";\n\nexport async function handle(req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.access_token}`, \"Content-Type\": \"application/json\" };\n const sp = new URLSearchParams(params);\n\n let url: string;\n switch (endpoint) {\n case \"events\":\n case \"upcoming\":\n if (!sp.has(\"timeMin\")) sp.set(\"timeMin\", new Date().toISOString());\n if (!sp.has(\"timeMax\")) sp.set(\"timeMax\", new Date(Date.now() + 30 * 86400000).toISOString());\n if (!sp.has(\"maxResults\")) sp.set(\"maxResults\", \"50\");\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n case \"calendars\":\n url = `${BASE}/users/me/calendarList?maxResults=250`;\n break;\n case \"today\": {\n const start = new Date(); start.setHours(0,0,0,0);\n const end = new Date(); end.setHours(23,59,59,999);\n sp.set(\"timeMin\", start.toISOString());\n sp.set(\"timeMax\", end.toISOString());\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n sp.set(\"maxResults\", \"50\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n }\n case \"week\": {\n const start = new Date(); start.setHours(0,0,0,0);\n const end = new Date(Date.now() + 7 * 86400000);\n sp.set(\"timeMin\", start.toISOString());\n sp.set(\"timeMax\", end.toISOString());\n sp.set(\"singleEvents\", \"true\");\n sp.set(\"orderBy\", \"startTime\");\n sp.set(\"maxResults\", \"100\");\n url = `${BASE}/calendars/primary/events?${sp.toString()}`;\n break;\n }\n case \"settings\":\n url = `${BASE}/users/me/settings`;\n break;\n // ─── WRITE ──────────────────────────────────────────────────────────────\n case \"createEvent\": {\n const body = await req.json().catch(() => ({}));\n if (!body.summary || !body.start || !body.end) return json({ error: \"summary, start, end required\" }, 400);\n const calId = body.calendarId || \"primary\";\n const payload: any = {\n summary: body.summary,\n description: body.description,\n location: body.location,\n start: { dateTime: new Date(body.start).toISOString() },\n end: { dateTime: new Date(body.end).toISOString() },\n };\n if (Array.isArray(body.attendees)) payload.attendees = body.attendees.map((email: string) => ({ email }));\n const res = await fetch(`${BASE}/calendars/${calId}/events?sendUpdates=all`, {\n method: \"POST\", headers, body: JSON.stringify(payload),\n });\n return json(await safeJson(res), res.status);\n }\n case \"deleteEvent\": {\n const body = await req.json().catch(() => ({}));\n if (!body.eventId) return json({ error: \"eventId required\" }, 400);\n const calId = body.calendarId || \"primary\";\n const res = await fetch(`${BASE}/calendars/${calId}/events/${body.eventId}`, { method: \"DELETE\", headers });\n if (res.status === 204) return json({ ok: true, deleted: body.eventId });\n return json(await safeJson(res), res.status);\n }\n default:\n url = `${BASE}/${endpoint}?${params}`;\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n",[],[365,917,680],"c3d36e94-5806-11f1-8bf1-ee650bb23af1","c3d36f5c-5806-11f1-8bf1-ee650bb23af1",112,["D",1779692150355],"// Ahrefs · SEO data (keywords, backlinks, top pages, domain rating, organic traffic).\n// Docs: https://docs.ahrefs.com/en/api/reference/site-explorer\n// Auth: Bearer api_key. Important quirks:\n// - /metrics, /domain-rating, /backlinks-stats do NOT accept a `select` parameter.\n// - /all-backlinks, /organic-keywords, /keywords-explorer/matching-terms REQUIRE `select`.\n// - The referring-domains endpoint path is /refdomains (not /referring-domains).\nimport { json, safeJson, IntegrationCredentials } from \"./_shared.ts\";\n\nconst BASE = \"https://api.ahrefs.com/v3\";\n\nexport async function handle(_req: Request, endpoint: string, params: string, creds: IntegrationCredentials): Promise<Response> {\n const headers = { Authorization: `Bearer ${creds.api_key}` };\n const today = new Date().toISOString().slice(0, 10);\n const sp = new URLSearchParams(params);\n const target = sp.get(\"target\") || creds.extra_config?.target || \"\";\n if (target) sp.set(\"target\", target);\n\n let url: string;\n switch (endpoint) {\n case \"site-explorer\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/metrics?${sp.toString()}`;\n break;\n }\n case \"keywords\": {\n if (!sp.has(\"select\")) sp.set(\"select\", \"keyword,volume,difficulty,cpc,traffic_potential,global_volume\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n const seed = sp.get(\"keywords\") || sp.get(\"keyword\") || sp.get(\"q\") || \"seo\";\n sp.delete(\"keyword\"); sp.delete(\"q\");\n sp.set(\"keywords\", seed);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/keywords-explorer/matching-terms?${sp.toString()}`;\n break;\n }\n case \"backlinks\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"url_from,domain_rating_source,anchor,name_source,title,first_seen,is_dofollow,traffic\");\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/all-backlinks?${sp.toString()}`;\n break;\n }\n case \"domain-rating\": {\n if (!sp.get(\"target\")) sp.set(\"target\", \"ahrefs.com\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/domain-rating?${sp.toString()}`;\n break;\n }\n case \"organic-keywords\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"keyword,volume,sum_traffic,best_position,cpc,keyword_difficulty\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/organic-keywords?${sp.toString()}`;\n break;\n }\n case \"top-pages\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"url,sum_traffic,keywords,top_keyword,top_keyword_best_position,value\");\n if (!sp.has(\"country\")) sp.set(\"country\", \"us\");\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/top-pages?${sp.toString()}`;\n break;\n }\n case \"referring-domains\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"select\")) sp.set(\"select\", \"domain,domain_rating,dofollow_links,dofollow_linked_domains,traffic_domain,first_seen\");\n if (!sp.has(\"limit\")) sp.set(\"limit\", \"50\");\n url = `${BASE}/site-explorer/refdomains?${sp.toString()}`;\n break;\n }\n case \"backlinks-stats\": {\n if (!sp.get(\"target\")) return json({ error: \"Target domain required. Set it in Ahrefs Settings.\" }, 400);\n if (!sp.has(\"date\")) sp.set(\"date\", today);\n sp.delete(\"select\");\n url = `${BASE}/site-explorer/backlinks-stats?${sp.toString()}`;\n break;\n }\n default:\n return json({ error: `Unknown Ahrefs endpoint: ${endpoint}` }, 400);\n }\n const res = await fetch(url, { headers });\n return json(await safeJson(res), res.status);\n}\n\nexport async function test(creds: IntegrationCredentials): Promise<{ ok: boolean; message: string }> {\n const today = new Date().toISOString().slice(0, 10);\n const res = await fetch(`${BASE}/site-explorer/domain-rating?target=ahrefs.com&date=${today}`, {\n headers: { Authorization: `Bearer ${creds.api_key}` },\n });\n if (res.status === 401 || res.status === 403) return { ok: false, message: \"Invalid API key\" };\n if (res.status === 429) return { ok: false, message: \"Rate limited — try again shortly\" };\n if (!res.ok) {\n const d = await safeJson(res);\n return { ok: false, message: d?.error ?? `API error ${res.status}` };\n }\n return { ok: true, message: \"API key verified\" };\n}\n",[],[365,917,688],"212361f0-5d69-11f1-bb77-1607ee4eb77e","_shared.ts","c3708540-5806-11f1-acd7-ee650bb23af1",612,["D",1780284153389],"// Common utilities shared across per-integration handler files.\n// Each integration handler file imports from here rather than duplicating these helpers.\nimport { sqlite } from \"../../db.ts\";\n\nexport interface IntegrationCredentials {\n api_key?: string;\n api_secret?: string;\n access_token?: string;\n refresh_token?: string;\n extra_config?: Record<string, any>;\n}\n\nconst CORS = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, PUT, DELETE, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n};\n\nexport function json(data: any, status = 200): Response {\n return new Response(JSON.stringify(data), {\n status,\n headers: { \"Content-Type\": \"application/json\", ...CORS },\n });\n}\n\nexport async function safeJson(res: Response): Promise<any> {\n const text = await res.text();\n try { return JSON.parse(text); } catch { return { error: text.slice(0, 500) || `HTTP ${res.status}` }; }\n}\n\n// SSRF guard for any server-side fetch built from user-supplied URLs (WordPress site_url,\n// custom LLM endpoint). Only allows public https hosts; blocks loopback/private/link-local\n// ranges and the cloud metadata IP. (Defense in depth — the APP_TOKEN gate already restricts\n// who can configure these.)\nexport function isSafePublicUrl(rawUrl: string): boolean {\n let u: URL;\n try { u = new URL(rawUrl); } catch { return false; }\n if (u.protocol !== \"https:\") return false;\n const host = u.hostname.toLowerCase().replace(/^\\[|\\]$/g, \"\");\n if (host === \"localhost\" || host.endsWith(\".localhost\") || host.endsWith(\".internal\") || host.endsWith(\".local\")) return false;\n if (host === \"::1\" || host === \"::\" || host.startsWith(\"fe80:\") || host.startsWith(\"fc\") || host.startsWith(\"fd\")) return false;\n const m = host.match(/^(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})$/);\n if (m) {\n const a = +m[1], b = +m[2];\n if (a === 0 || a === 127) return false; // unspecified / loopback\n if (a === 10) return false; // 10/8 private\n if (a === 172 && b >= 16 && b <= 31) return false; // 172.16/12 private\n if (a === 192 && b === 168) return false; // 192.168/16 private\n if (a === 169 && b === 254) return false; // link-local / cloud metadata\n }\n return true;\n}\n\nexport async function getCredentials(name: string): Promise<IntegrationCredentials | null> {\n const result = await sqlite.execute(\n \"SELECT api_key, api_secret, access_token, refresh_token, extra_config FROM integrations WHERE name = ?\",\n [name]\n );\n if (!result.rows || result.rows.length === 0) {\n // Google-family services (drive/sheets/docs/gmail/etc.) share the central 'google' OAuth\n // credentials and usually have NO row of their own. Inherit the google creds so the router\n // doesn't 404 before withGoogleFallback runs. ('google' is not in GOOGLE_SERVICES → no recursion.)\n if (GOOGLE_SERVICES.includes(name)) return await getCredentials(\"google\");\n // Microsoft-family services (onedrive/teams/excel) share the central 'outlook' Graph token.\n if (MICROSOFT_SERVICES.includes(name)) return await getCredentials(\"outlook\");\n return null;\n }\n const cols = result.columns;\n const row = result.rows[0];\n const obj: Record<string, any> = {};\n for (let i = 0; i < cols.length; i++) obj[cols[i]] = row[i];\n let extra_config: Record<string, any> | undefined;\n if (obj.extra_config) {\n try { extra_config = JSON.parse(obj.extra_config); } catch { extra_config = {}; }\n }\n return {\n api_key: obj.api_key ?? undefined,\n api_secret: obj.api_secret ?? undefined,\n access_token: obj.access_token ?? undefined,\n refresh_token: obj.refresh_token ?? undefined,\n extra_config,\n };\n}\n\n// All Google services share the OAuth token issued for `google`.\nexport async function withGoogleFallback(creds: IntegrationCredentials): Promise<IntegrationCredentials> {\n const google = await getCredentials(\"google\");\n return {\n ...creds,\n access_token: creds.access_token || google?.access_token,\n refresh_token: creds.refresh_token || google?.refresh_token,\n api_key: creds.api_key || google?.api_key,\n extra_config: { ...google?.extra_config, ...creds.extra_config },\n };\n}\n\n// Refresh Google access_token using stored refresh_token. Persists the new token and returns it.\nexport async function refreshGoogleToken(): Promise<string | null> {\n const google = await getCredentials(\"google\");\n if (!google?.refresh_token) return null;\n // Match the callback handler: env-driven OAuth client by design, DB row as fallback. Trim to\n // avoid whitespace producing a malformed (invalid_request) token request.\n const clientId = (Deno.env.get(\"GOOGLE_CLIENT_ID\") || google.extra_config?.client_id || google.extra_config?.clientId || \"\").trim();\n const clientSecret = (Deno.env.get(\"GOOGLE_CLIENT_SECRET\") || google.api_secret || \"\").trim();\n if (!clientId || !clientSecret) return null;\n const res = await fetch(\"https://oauth2.googleapis.com/token\", {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({ refresh_token: google.refresh_token, client_id: clientId, client_secret: clientSecret, grant_type: \"refresh_token\" }),\n });\n const data = await res.json();\n if (!data.access_token) return null;\n await sqlite.execute(\"UPDATE integrations SET access_token=?, updated_at=? WHERE name='google'\", [data.access_token, new Date().toISOString()]);\n return data.access_token;\n}\n\n// List of integrations that share the central Google OAuth credentials.\nexport const GOOGLE_SERVICES = [\"ga4\", \"gsc\", \"youtube\", \"drive\", \"sheets\", \"calendar\", \"gmail\", \"docs\"];\n\n// Env-driven Google OAuth client helpers. The OAuth client id/secret live in the\n// GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET env vars by design. Re-exported by api/integrations.ts;\n// removing these breaks that module's import graph (and the whole app).\nexport const googleClientId = (): string => (Deno.env.get(\"GOOGLE_CLIENT_ID\") || \"\").trim();\nexport const googleClientSecret = (): string => (Deno.env.get(\"GOOGLE_CLIENT_SECRET\") || \"\").trim();\nexport const googleRedirectUri = (): string => (Deno.env.get(\"GOOGLE_REDIRECT_URI\") || \"\").trim();\n\n// ─── Microsoft (Outlook / Graph) OAuth ──────────────────────────────────────\n// Env-driven confidential client, same pattern as Google. `common` tenant accepts both\n// work/school and personal Microsoft accounts. Delegated scopes cover the Outlook app's\n// read (inbox/calendar/contacts) + write (send) endpoints; offline_access yields a refresh token.\nexport const MICROSOFT_TENANT = \"common\";\nexport const MICROSOFT_SCOPES =\n \"openid profile email offline_access User.Read Mail.Read Mail.Send Calendars.Read Contacts.Read \" +\n \"Files.Read.All Files.ReadWrite.All Team.ReadBasic.All Channel.ReadBasic.All ChannelMessage.Read.All \" +\n \"ChannelMessage.Send Chat.ReadWrite ChatMessage.Send OnlineMeetings.ReadWrite\";\nexport const microsoftClientId = (): string => (Deno.env.get(\"MICROSOFT_CLIENT_ID\") || \"\").trim();\nexport const microsoftClientSecret = (): string => (Deno.env.get(\"MICROSOFT_CLIENT_SECRET\") || \"\").trim();\nexport const microsoftTokenUrl = (): string =>\n `https://login.microsoftonline.com/${MICROSOFT_TENANT}/oauth2/v2.0/token`;\n\n// Refresh the Outlook access_token using its stored refresh_token. Persists the new token\n// (and the rotated refresh_token, if Microsoft returns one) and returns the new access token.\nexport async function refreshMicrosoftToken(): Promise<string | null> {\n const outlook = await getCredentials(\"outlook\");\n if (!outlook?.refresh_token) return null;\n // Env-first, falling back to the stored outlook credential row (same as the OAuth routes).\n const clientId = (microsoftClientId() || outlook.extra_config?.client_id || outlook.extra_config?.clientId || \"\").trim();\n const clientSecret = (microsoftClientSecret() || outlook.api_secret || \"\").trim();\n if (!clientId || !clientSecret) return null;\n const res = await fetch(microsoftTokenUrl(), {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/x-www-form-urlencoded\" },\n body: new URLSearchParams({\n refresh_token: outlook.refresh_token,\n client_id: clientId,\n client_secret: clientSecret,\n grant_type: \"refresh_token\",\n scope: MICROSOFT_SCOPES,\n }),\n });\n const data = await res.json();\n if (!data.access_token) return null;\n await sqlite.execute(\n \"UPDATE integrations SET access_token=?, refresh_token=COALESCE(?, refresh_token), updated_at=? WHERE name='outlook'\",\n [data.access_token, data.refresh_token ?? null, new Date().toISOString()],\n );\n return data.access_token;\n}\n\n// Microsoft Graph services that share the central Outlook OAuth token (mirror GOOGLE_SERVICES).\nexport const MICROSOFT_SERVICES = [\"onedrive\", \"teams\", \"excel\"];\n\n// All Microsoft services share the OAuth token issued for `outlook` (the Graph token).\nexport async function withMicrosoftFallback(creds: IntegrationCredentials): Promise<IntegrationCredentials> {\n const outlook = await getCredentials(\"outlook\");\n return {\n ...creds,\n access_token: creds.access_token || outlook?.access_token,\n refresh_token: creds.refresh_token || outlook?.refresh_token,\n api_secret: creds.api_secret || outlook?.api_secret,\n extra_config: { ...outlook?.extra_config, ...creds.extra_config },\n };\n}\n",[],[365,917,1025],"c370b894-5806-11f1-acd7-ee650bb23af1",111,["D",1779692149653],[382,383,384,385,386,387,388,389,390,391,392,393,394,395,396,397,398],[365,917],["D",1779556136787],[381,399,400,401,402,403,404,405,406,407,408,409,410,411,412,413,414,415,416,417,418,419,420,421,422,423],[365],["D",1779556068488],["D",1789503703835],{"_99":98,"_106":102},{"_99":98,"_100":98,"_101":102,"_18":54,"_103":104,"_105":24},{"_13":88,"_31":89,"_42":41,"_90":43,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":44,"_97":98},"ctx",{"_81":-5,"_126":1078,"_1079":1080,"_1081":1082,"_1083":24,"_1084":24,"_1085":24,"_1086":58,"_1087":-5,"_1088":-7,"_1089":-7,"_1090":62,"_1091":24,"_1092":24,"_1093":-5,"_1094":24,"_1095":24,"_1096":24,"_1097":-5},"ENV",{"_1049":1050,"_1051":1052,"_1053":89,"_1054":1055,"_1056":1057,"_1058":1059,"_1060":1061,"_1062":1063,"_1064":1065,"_1066":1067,"_1068":1069,"_1070":1071,"_1072":1073,"_1074":1075,"_1076":1077},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"] </article></body></html>

Diff · v645