Hi

\",\n});\n```\n\n## Val Town Utility Functions\n\nVal Town provides several utility functions to help with common project tasks.\n\n### Importing Utilities\n\nAlways import utilities with version pins to avoid breaking changes:\n\n```ts\nimport {\n parseProject,\n readFile,\n serveFile,\n} from \"https://esm.town/v/std/utils@85-main/index.ts\";\n```\n\n### Available Utilities\n\n#### **serveFile** - Serve project files with proper content types\n\nFor example, in Hono:\n\n```ts\n// serve all files in frontend/ and shared/\napp.get(\"/frontend/*\", (c) => serveFile(c.req.path, import.meta.url));\napp.get(\"/shared/*\", (c) => serveFile(c.req.path, import.meta.url));\n```\n\n#### **readFile** - Read files from within the project:\n\n```ts\n// Read a file from the project\nconst fileContent = await readFile(\"/frontend/index.html\", import.meta.url);\n```\n\n#### **listFiles** - List all files in the project\n\n```ts\nconst files = await listFiles(import.meta.url);\n```\n\n#### **parseProject** - Extract information about the current project from import.meta.url\n\nThis is useful for including info for linking back to a val, ie in \"view source\"\nurls:\n\n```ts\nconst projectVal = parseProject(import.meta.url);\nconsole.log(projectVal.username); // Owner of the project\nconsole.log(projectVal.name); // Project name\nconsole.log(projectVal.version); // Version number\nconsole.log(projectVal.branch); // Branch name\nconsole.log(projectVal.links.self.project); // URL to the project page\n```\n\nHowever, it's _extremely importing_ to note that `parseProject` and other\nStandard Library utilities ONLY RUN ON THE SERVER. If you need access to this\ndata on the client, run it in the server and pass it to the client by splicing\nit into the HTML page or by making an API request for it.\n\n## Val Town Platform Specifics\n\n- **Redirects:** Use\n `return new Response(null, { status: 302, headers: { Location: \"/place/to/redirect\" }})`\n instead of `Response.redirect` which is broken\n- **Images:** Avoid external images or base64 images. Use emojis, unicode\n symbols, or icon fonts/libraries instead\n- **AI Image:** To inline generate an AI image use:\n `

`\n- **Storage:** DO NOT use the Deno KV module for storage\n- **Browser APIs:** DO NOT use the `alert()`, `prompt()`, or `confirm()` methods\n- **Weather Data:** Use open-meteo for weather data (doesn't require API keys)\n unless otherwise specified\n- **View Source:** Add a view source link by importing & using\n `import.meta.url.replace(\"ems.sh\", \"val.town)\"` (or passing this data to the\n client) and include `target=\"_top\"` attribute\n- **Error Debugging:** Add\n `` to HTML to capture\n client-side errors\n- **Error Handling:** Only use try...catch when there's a clear local\n resolution; Avoid catches that merely log or return 500s. Let errors bubble up\n with full context\n- **Environment Variables:** Use `Deno.env.get('keyname')` when you need to, but\n generally prefer APIs that don't require keys\n- **Imports:** Use `https://esm.sh` for npm and Deno dependencies to ensure\n compatibility on server and browser\n- **Storage Strategy:** Only use backend storage if explicitly required; prefer\n simple static client-side sites\n- **React Configuration:** When using React libraries, pin versions with\n `?deps=react@18.2.0,react-dom@18.2.0` and start the file with\n `/** @jsxImportSource https://esm.sh/react@18.2.0 */`\n- Ensure all React dependencies and sub-dependencies are pinned to the same\n version\n- **Styling:** Default to using TailwindCSS via\n `` unless otherwise\n specified\n\n## Project Structure and Design Patterns\n\n### Recommended Directory Structure\n\n```\n├── backend/\n│ ├── database/\n│ │ ├── migrations.ts # Schema definitions\n│ │ ├── queries.ts # DB query functions\n│ │ └── README.md\n│ └── routes/ # Route modules\n│ ├── [route].ts\n│ └── static.ts # Static file serving\n│ ├── index.ts # Main entry point\n│ └── README.md\n├── frontend/\n│ ├── components/\n│ │ ├── App.tsx\n│ │ └── [Component].tsx\n│ ├── favicon.svg\n│ ├── index.html # Main HTML template\n│ ├── index.tsx # Frontend JS entry point\n│ ├── README.md\n│ └── style.css\n├── README.md\n└── shared/\n ├── README.md\n └── utils.ts # Shared types and functions\n```\n\n### Backend (Hono) Best Practices\n\n- Hono is the recommended API framework\n- Main entry point should be `backend/index.ts`\n- **Static asset serving:** Use the utility functions to read and serve project\n files:\n ```ts\n import {\n readFile,\n serveFile,\n } from \"https://esm.town/v/std/utils@85-main/index.ts\";\n\n // serve all files in frontend/ and shared/\n app.get(\"/frontend/*\", (c) => serveFile(c.req.path, import.meta.url));\n app.get(\"/shared/*\", (c) => serveFile(c.req.path, import.meta.url));\n\n // For index.html, often you'll want to bootstrap with initial data\n app.get(\"/\", async (c) => {\n let html = await readFile(\"/frontend/index.html\", import.meta.url);\n\n // Inject data to avoid extra round-trips\n const initialData = await fetchInitialData();\n const dataScript = ``;\n\n html = html.replace(\"\", `${dataScript}`);\n return c.html(html);\n });\n ```\n- Create RESTful API routes for CRUD operations\n- Always include this snippet at the top-level Hono app to re-throwing errors to\n see full stack traces:\n ```ts\n // Unwrap Hono errors to see original error details\n app.onError((err, c) => {\n throw err;\n });\n ```\n\n### Database Patterns\n\n- Run migrations on startup or comment out for performance\n- Change table names when modifying schemas rather than altering\n- Export clear query functions with proper TypeScript typing\n\n## Common Gotchas and Solutions\n\n1. **Environment Limitations:**\n - Val Town runs on Deno in a serverless context, not Node.js\n - Code in `shared/` must work in both frontend and backend environments\n - Cannot use `Deno` keyword in shared code\n - Use `https://esm.sh` for imports that work in both environments\n\n2. **SQLite Peculiarities:**\n - Limited support for ALTER TABLE operations\n - Create new tables with updated schemas and copy data when needed\n - Always run table creation before querying\n\n3. **React Configuration:**\n - All React dependencies must be pinned to 18.2.0\n - Always include `@jsxImportSource https://esm.sh/react@18.2.0` at the top of\n React files\n - Rendering issues often come from mismatched React versions\n\n4. **File Handling:**\n - Val Town only supports text files, not binary\n - Use the provided utilities to read files across branches and forks\n - For files in the project, use `readFile` helpers\n\n5. **API Design:**\n - `fetch` handler is the entry point for HTTP vals\n - Run the Hono app with\n `export default app.fetch // This is the entry point for HTTP vals`",[],[132],"0cfd9aca-0a11-11f1-8914-42dde27851f2",".vtignore","019c5f00-3df9-72b9-8e7f-320cf170942a",["D",1771120376884],".git\n.vscode\n.cursorrules\n.DS_Store\nnode_modules\nvendor",[],[139],["D",1771120376884],[148],[19],{"_36":60,"_13":61,"_62":63,"_64":65,"_66":67,"_68":34,"_69":70,"_38":149,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":150,"_18":151},["D",1771120376884],[152,153,154,155,156],[19,61],{"_36":185,"_13":186,"_62":187,"_64":63,"_66":67,"_68":34,"_69":70,"_38":188,"_17":32,"_72":97,"_74":189,"_75":51,"_76":-5,"_99":190,"_18":191},{"_36":178,"_13":179,"_62":180,"_64":63,"_66":67,"_68":34,"_69":70,"_38":181,"_17":32,"_72":97,"_74":182,"_75":51,"_76":-5,"_99":183,"_18":184},{"_36":171,"_13":172,"_62":173,"_64":63,"_66":67,"_68":34,"_69":70,"_38":174,"_17":32,"_72":97,"_74":175,"_75":51,"_76":-5,"_99":176,"_18":177},{"_36":164,"_13":165,"_62":166,"_64":63,"_66":67,"_68":34,"_69":70,"_38":167,"_17":32,"_72":97,"_74":168,"_75":51,"_76":-5,"_99":169,"_18":170},{"_36":157,"_13":158,"_62":159,"_64":63,"_66":67,"_68":34,"_69":70,"_38":160,"_17":32,"_72":97,"_74":161,"_75":51,"_76":-5,"_99":162,"_18":163},"0cfd8ec2-0a11-11f1-8914-42dde27851f2","types.ts","019c5f00-3df9-72b9-8e7f-2470bb552f5b",["D",1771120376884],"/**\n * Shared types for skills indexing\n */\n\n/**\n * Input reference to a SKILL.md file to be indexed\n */\nexport type SkillFileRef = {\n repoFullName: string; // \"owner/repo\"\n repoHtmlUrl: string; // \"https://github.com/owner/repo\"\n path: string; // \"skills/foo/SKILL.md\"\n};\n\n/**\n * Status of a single skill after processing\n */\nexport type SkillProcessingStatus =\n | \"indexed\"\n | \"unchanged\"\n | \"invalid\"\n | \"skipped\"\n | \"failed\";\n\n/**\n * Result for a single processed skill\n */\nexport type ProcessedSkillResult = {\n namespace: string;\n repoFullName: string;\n path: string;\n status: SkillProcessingStatus;\n reason?: string;\n skillName?: string;\n name?: string;\n description?: string;\n relDir?: string;\n sourceUrl?: string;\n};\n\n/**\n * Result of indexing a batch of skill files\n */\nexport type IndexSkillFilesResult = {\n found: number;\n indexed: number;\n invalid: number;\n failed: number;\n skipped: number;\n unchanged: number;\n duplicatesMarked: number;\n skills: ProcessedSkillResult[];\n};\n\n/**\n * Internal type for a skill being processed through the pipeline\n */\nexport type ProcessedSkill = {\n namespace: string;\n contentHash: string;\n firstCommitAt: string | null;\n dbSkill: {\n id: string;\n name: string;\n namespace: string;\n sourceUrl: string;\n sourceType: string;\n description: string;\n version: string | null;\n dependencies: string | null;\n author: string;\n stars: number;\n installs: number;\n verified: boolean;\n metadata: string;\n createdAt: string;\n updatedAt: string;\n firstCommitAt: string | null;\n contentHash: string | null;\n };\n repoOwner: string;\n repoName: string;\n repoFullName: string;\n path: string;\n};\n",[],[19,61,158],"0cfd7c5c-0a11-11f1-8914-42dde27851f2","skill-parser.ts","019c5f00-3df9-72b9-8e7f-23d06ebe6636",["D",1771120376884],"import { z } from \"npm:zod@4.1.12\";\n\n/**\n * Zod schema for SKILL.md YAML frontmatter\n * Based on Claude Code Skills documentation\n */\nconst SkillFrontmatterSchema = z\n .object({\n name: z.string().min(1).max(64), // Required, max 64 chars\n description: z.string().min(1),\n version: z.string().optional(),\n dependencies: z.string().optional(),\n })\n .passthrough(); // Allow other fields\n\nexport type SkillFrontmatter = z.infer;\n\n/**\n * Parse and validate SKILL.md content\n * Extracts YAML frontmatter and validates required fields\n */\nexport function parseSkillMd(\n content: string,\n repoFullName: string,\n): SkillFrontmatter | null {\n try {\n // Extract YAML frontmatter\n const frontmatterMatch = content.match(/^---\\n([\\s\\S]*?)\\n---/);\n\n if (!frontmatterMatch) {\n console.warn(\n `No YAML frontmatter found in SKILL.md from ${repoFullName}`,\n );\n return null;\n }\n\n const yamlContent = frontmatterMatch[1];\n\n // Simple YAML parser (handles basic key: value pairs)\n const frontmatter: Record = {};\n const lines = yamlContent.split(\"\\n\");\n\n for (const line of lines) {\n const match = line.match(/^(\\w+):\\s*(.+)$/);\n if (match) {\n const [, key, value] = match;\n // Remove quotes if present\n frontmatter[key] = value.replace(/^[\"']|[\"']$/g, \"\").trim();\n }\n }\n\n // Validate with Zod\n const result = SkillFrontmatterSchema.safeParse(frontmatter);\n\n if (!result.success) {\n console.error(`Invalid SKILL.md frontmatter from ${repoFullName}:`);\n console.error(\n `Validation errors: ${JSON.stringify(result.error.issues, null, 2)}`,\n );\n return null;\n }\n\n return result.data;\n } catch (error) {\n console.error(\n `Failed to parse SKILL.md from ${repoFullName}:`,\n error instanceof Error ? error.message : String(error),\n );\n return null;\n }\n}\n\n/**\n * Transform skill data into database record\n */\nexport function transformToDbSkill(\n skillData: SkillFrontmatter,\n repoOwner: string,\n repoName: string,\n directoryPath: string,\n repoUrl: string,\n repoStars: number,\n rawFileUrl: string,\n defaultBranch: string = \"main\",\n firstCommitAt: string | null = null,\n contentHash: string | null = null,\n): {\n id: string;\n name: string;\n namespace: string;\n sourceUrl: string;\n sourceType: string;\n description: string;\n version: string | null;\n dependencies: string | null;\n author: string;\n stars: number;\n installs: number;\n verified: boolean;\n metadata: string;\n createdAt: string;\n updatedAt: string;\n firstCommitAt: string | null;\n contentHash: string | null;\n} {\n // Extract skill directory name from path\n const skillDirName = directoryPath.split(\"/\").filter(Boolean).pop() ||\n \"unknown\";\n\n // Build namespace: @owner/repo/skillName\n const namespace = `@${repoOwner}/${repoName}/${skillDirName}`;\n\n // Build source URL (GitHub tree URL to skill directory)\n const sourceUrl = `${repoUrl}/tree/${defaultBranch}/${directoryPath}`;\n\n // Metadata blob\n const metadata = {\n repoOwner,\n repoName,\n directoryPath,\n rawFileUrl,\n };\n\n return {\n id: crypto.randomUUID(),\n name: skillData.name,\n namespace,\n sourceUrl,\n sourceType: \"github\",\n description: skillData.description,\n version: skillData.version || null,\n dependencies: skillData.dependencies || null,\n author: repoOwner,\n stars: repoStars,\n installs: 0,\n verified: false,\n metadata: JSON.stringify(metadata),\n createdAt: new Date().toISOString(),\n updatedAt: new Date().toISOString(),\n firstCommitAt,\n contentHash,\n };\n}",[],[19,61,165],"0cfd5312-0a11-11f1-8914-42dde27851f2","index-skill-files.ts","019c5f00-3df9-72b9-8e7f-1e0871216a3a",["D",1771120376884],"import { parseSkillMd, transformToDbSkill } from \"./skill-parser.ts\";\nimport type {\n IndexSkillFilesResult,\n ProcessedSkill,\n SkillFileRef,\n} from \"./types.ts\";\nimport { GitHubClient } from \"../../clients/github.ts\";\nimport { db } from \"../../database/db.ts\";\nimport { skills } from \"../../database/schema.ts\";\nimport { inArray, sql } from \"npm:drizzle-orm@0.44.2\";\nimport { findDuplicates } from \"./find-duplicates.ts\";\nimport { chunkArray, computeHash } from \"../../utils.ts\";\n\nexport type Logger = {\n info: (message: string, data?: Record) => void;\n warn: (message: string, data?: Record) => void;\n error: (message: string, data?: Record) => void;\n};\n\nexport type IndexSkillFilesOptions = {\n github: GitHubClient;\n log: Logger;\n dryRun?: boolean;\n};\n\nexport async function indexSkillFiles(\n refs: SkillFileRef[],\n { github, log, dryRun = false }: IndexSkillFilesOptions,\n): Promise {\n const result: IndexSkillFilesResult = {\n found: refs.length,\n indexed: 0,\n invalid: 0,\n failed: 0,\n skipped: 0,\n unchanged: 0,\n duplicatesMarked: 0,\n skills: [],\n };\n\n if (refs.length === 0) {\n log.info(\"No skill files to process\");\n return result;\n }\n\n log.info(`Processing ${refs.length} skill files...`);\n\n const allProcessedSkills: ProcessedSkill[] = [];\n\n const processResults = await Promise.allSettled(\n refs.map(async (ref): Promise<{\n status: \"success\" | \"invalid\" | \"skipped\";\n reason?: string;\n ref: SkillFileRef;\n data?: ProcessedSkill;\n }> => {\n const content = await github.fetchFileContent(ref.repoFullName, ref.path);\n const contentHash = computeHash(content);\n const skillData = parseSkillMd(content, ref.repoFullName);\n\n if (!contentHash || !skillData) {\n return { status: \"invalid\", reason: \"Invalid SKILL.md frontmatter\", ref };\n }\n\n const repoMetadata = await github.fetchRepoMetadata(ref.repoFullName);\n if (repoMetadata.fork) {\n return { status: \"skipped\", reason: \"fork\", ref };\n }\n\n const directoryPath = ref.path.replace(/\\/SKILL\\.md$/, \"\");\n const rawFileUrl = `https://raw.githubusercontent.com/${ref.repoFullName}/${repoMetadata.default_branch}/${ref.path}`;\n const [repoOwner, repoName] = ref.repoFullName.split(\"/\");\n\n const dbSkill = transformToDbSkill(\n skillData,\n repoOwner,\n repoName,\n directoryPath,\n ref.repoHtmlUrl,\n repoMetadata.stargazers_count,\n rawFileUrl,\n repoMetadata.default_branch,\n null,\n contentHash,\n );\n\n return {\n status: \"success\",\n ref,\n data: {\n namespace: dbSkill.namespace,\n contentHash,\n firstCommitAt: null,\n dbSkill,\n repoOwner,\n repoName,\n repoFullName: ref.repoFullName,\n path: ref.path,\n },\n };\n }),\n );\n\n for (const processResult of processResults) {\n if (processResult.status === \"rejected\") {\n log.error(\"Failed to process skill file\", { error: String(processResult.reason) });\n result.failed++;\n continue;\n }\n\n const value = processResult.value;\n\n if (value.status === \"invalid\") {\n log.warn(`Invalid SKILL.md from ${value.ref.repoFullName} ${value.ref.path}`);\n result.invalid++;\n result.skills.push({\n namespace: `@${value.ref.repoFullName}/${value.ref.path.replace(/\\/SKILL\\.md$/, \"\").split(\"/\").pop()}`,\n repoFullName: value.ref.repoFullName,\n path: value.ref.path,\n status: \"invalid\",\n reason: value.reason,\n });\n continue;\n }\n\n if (value.status === \"skipped\") {\n log.info(`Skipped skill (${value.reason}): ${value.ref.repoFullName} ${value.ref.path}`);\n result.skipped++;\n result.skills.push({\n namespace: `@${value.ref.repoFullName}/${value.ref.path.replace(/\\/SKILL\\.md$/, \"\").split(\"/\").pop()}`,\n repoFullName: value.ref.repoFullName,\n path: value.ref.path,\n status: \"skipped\",\n reason: value.reason,\n });\n continue;\n }\n\n if (value.data) {\n allProcessedSkills.push(value.data);\n }\n }\n\n if (allProcessedSkills.length === 0) {\n log.info(\"No valid skills to index\");\n return result;\n }\n\n log.info(`Collected ${allProcessedSkills.length} valid skills for batch processing`);\n\n const namespaces = allProcessedSkills.map((s) => s.namespace);\n const existingSkillsMap = new Map<\n string,\n { id: string; firstCommitAt: string | null; contentHash: string | null }\n >();\n\n for (const namespaceChunk of chunkArray(namespaces, 500)) {\n const existingSkills = await db\n .select({\n id: skills.id,\n namespace: skills.namespace,\n firstCommitAt: skills.firstCommitAt,\n contentHash: skills.contentHash,\n })\n .from(skills)\n .where(inArray(skills.namespace, namespaceChunk));\n\n for (const skill of existingSkills) {\n existingSkillsMap.set(skill.namespace, {\n id: skill.id,\n firstCommitAt: skill.firstCommitAt,\n contentHash: skill.contentHash,\n });\n }\n }\n\n log.info(`Found ${existingSkillsMap.size} existing skills in DB`);\n\n for (const skill of allProcessedSkills) {\n const existing = existingSkillsMap.get(skill.namespace);\n if (existing?.firstCommitAt) {\n skill.firstCommitAt = existing.firstCommitAt;\n }\n }\n\n const skillsNeedingFirstCommit = allProcessedSkills.filter((s) => !s.firstCommitAt);\n\n if (skillsNeedingFirstCommit.length > 0) {\n log.info(`Fetching firstCommitAt for ${skillsNeedingFirstCommit.length} skills`);\n\n const skillsByNamespace = new Map();\n for (const skill of allProcessedSkills) {\n skillsByNamespace.set(skill.namespace, skill);\n }\n\n for (const chunk of chunkArray(skillsNeedingFirstCommit, 50)) {\n const commitResults = await Promise.allSettled(\n chunk.map(async (skill) => {\n const firstCommitAt = await github.fetchFirstCommitDate(skill.repoFullName, skill.path);\n return { namespace: skill.namespace, firstCommitAt };\n }),\n );\n\n for (const commitResult of commitResults) {\n if (commitResult.status === \"fulfilled\" && commitResult.value.firstCommitAt) {\n const skill = skillsByNamespace.get(commitResult.value.namespace);\n if (skill) {\n skill.firstCommitAt = commitResult.value.firstCommitAt;\n }\n }\n }\n }\n }\n\n const skillsWithDates = allProcessedSkills\n .filter((s) => s.firstCommitAt)\n .map((s) => ({\n namespace: s.namespace,\n contentHash: s.contentHash,\n firstCommitAt: new Date(s.firstCommitAt!),\n }));\n\n const duplicateNamespaces = await findDuplicates(skillsWithDates);\n result.duplicatesMarked = duplicateNamespaces.size;\n log.info(`Found ${duplicateNamespaces.size} duplicates`);\n\n const skillRecords = allProcessedSkills\n .filter((skill) => {\n const existing = existingSkillsMap.get(skill.namespace);\n return !existing || existing.contentHash !== skill.contentHash;\n })\n .map((skill) => {\n const existing = existingSkillsMap.get(skill.namespace);\n const isDuplicate = duplicateNamespaces.has(skill.namespace);\n\n return {\n ...skill.dbSkill,\n id: existing?.id ?? skill.dbSkill.id,\n firstCommitAt: skill.firstCommitAt,\n deleted: false,\n duplicate: isDuplicate,\n };\n });\n\n const unchangedCount = allProcessedSkills.length - skillRecords.length;\n result.unchanged = unchangedCount;\n log.info(`Skipped ${unchangedCount} unchanged skills`);\n\n for (const skill of allProcessedSkills) {\n const existing = existingSkillsMap.get(skill.namespace);\n const isUnchanged = existing && existing.contentHash === skill.contentHash;\n const relDir = skill.path.replace(/\\/SKILL\\.md$/, \"\");\n const skillName = relDir === \"\" ? skill.repoName : relDir.split(\"/\").pop() || skill.repoName;\n\n result.skills.push({\n namespace: skill.namespace,\n repoFullName: skill.repoFullName,\n path: skill.path,\n status: isUnchanged ? \"unchanged\" : \"indexed\",\n skillName,\n name: skill.dbSkill.name,\n description: skill.dbSkill.description,\n relDir: relDir || \".\",\n sourceUrl: skill.dbSkill.sourceUrl,\n });\n }\n\n if (dryRun) {\n log.info(`Dry run: would upsert ${skillRecords.length} skills`);\n result.indexed = skillRecords.length;\n return result;\n }\n\n for (const chunk of chunkArray(skillRecords, 100)) {\n try {\n await db\n .insert(skills)\n .values(chunk)\n .onConflictDoUpdate({\n target: [skills.namespace],\n set: {\n name: sql`excluded.name`,\n description: sql`excluded.description`,\n version: sql`excluded.version`,\n dependencies: sql`excluded.dependencies`,\n sourceUrl: sql`excluded.source_url`,\n stars: sql`excluded.stars`,\n metadata: sql`excluded.metadata`,\n updatedAt: sql`CURRENT_TIMESTAMP`,\n firstCommitAt: sql`COALESCE(excluded.first_commit_at, skills.first_commit_at)`,\n contentHash: sql`excluded.content_hash`,\n duplicate: sql`excluded.duplicate`,\n },\n })\n .run();\n\n result.indexed += chunk.length;\n } catch (dbError) {\n log.error(`Failed to upsert skill batch`, { error: String(dbError) });\n result.failed += chunk.length;\n\n for (const record of chunk) {\n const skillResult = result.skills.find((s) => s.namespace === record.namespace);\n if (skillResult) {\n skillResult.status = \"failed\";\n skillResult.reason = \"Database error\";\n }\n }\n }\n }\n\n log.info(`Upserted ${result.indexed} skills`);\n\n return result;\n}",[],[19,61,172],"0cfd4b88-0a11-11f1-8914-42dde27851f2","find-duplicates.ts","019c5f00-3df9-72b9-8e7f-1ba3dd4c45b1",["D",1771120376884],"import { db } from \"../../database/db.ts\";\nimport { skills } from \"../../database/schema.ts\";\nimport { inArray } from \"npm:drizzle-orm@0.44.2\";\nimport { chunkArray } from \"../../utils.ts\";\n\nexport async function findDuplicates(\n skillsToCheck: Array<\n { namespace: string; contentHash: string; firstCommitAt: Date }\n >,\n): Promise> {\n if (skillsToCheck.length === 0) return new Set();\n\n const hashes = skillsToCheck.map((s) => s.contentHash);\n const duplicateNamespaces = new Set();\n\n const priorSkills: Array<{\n contentHash: string | null;\n namespace: string;\n firstCommitAt: string | null;\n }> = [];\n\n for (const hashChunk of chunkArray(hashes, 500)) {\n const results = await db\n .select({\n contentHash: skills.contentHash,\n namespace: skills.namespace,\n firstCommitAt: skills.firstCommitAt,\n })\n .from(skills)\n .where(inArray(skills.contentHash, hashChunk));\n\n priorSkills.push(...results);\n }\n\n for (const skill of skillsToCheck) {\n const hasPrior = priorSkills.some((prior) => {\n if (!prior.firstCommitAt) return false;\n return (\n prior.contentHash === skill.contentHash &&\n prior.namespace !== skill.namespace &&\n new Date(prior.firstCommitAt) < skill.firstCommitAt\n );\n });\n\n if (hasPrior) {\n duplicateNamespaces.add(skill.namespace);\n }\n }\n\n return duplicateNamespaces;\n}\n",[],[19,61,179],"0cfd45ca-0a11-11f1-8914-42dde27851f2","discover.ts","019c5f00-3df9-72b9-8e7f-14b4733bec51",["D",1771120376884],"import { GitHubClient, ListRepoFilesResult } from \"../../clients/github.ts\";\nimport type { SkillFileRef } from \"./types.ts\";\n\nexport type DiscoverSkillsResult = {\n repoFullName: string;\n repoHtmlUrl: string;\n defaultBranch: string;\n skills: SkillFileRef[];\n truncated: boolean;\n};\n\nexport function filterSkillFiles(\n repoFiles: ListRepoFilesResult,\n): SkillFileRef[] {\n return repoFiles.files\n .filter((f) => f.type === \"blob\" && f.path.endsWith(\"SKILL.md\"))\n .map((f) => ({\n repoFullName: repoFiles.repoFullName,\n repoHtmlUrl: repoFiles.repoHtmlUrl,\n path: f.path,\n }));\n}\n\nexport async function discoverSkillsInRepo(\n github: GitHubClient,\n repoFullName: string,\n): Promise {\n const repoFiles = await github.listRepoFiles(repoFullName);\n const skills = filterSkillFiles(repoFiles);\n\n return {\n repoFullName: repoFiles.repoFullName,\n repoHtmlUrl: repoFiles.repoHtmlUrl,\n defaultBranch: repoFiles.defaultBranch,\n skills,\n truncated: repoFiles.truncated,\n };\n}\n",[],[19,61,186],"0cfd4430-0a11-11f1-8914-42dde27851f2","database","019c5f00-3df9-72b9-8e7f-1269a3603ef5",["D",1771120376884],[198,199,200],[193],{"_36":215,"_13":216,"_62":217,"_64":194,"_66":67,"_68":34,"_69":70,"_38":218,"_17":32,"_72":97,"_74":219,"_75":51,"_76":-5,"_99":220,"_18":221},{"_36":208,"_13":209,"_62":210,"_64":194,"_66":67,"_68":34,"_69":70,"_38":211,"_17":32,"_72":97,"_74":212,"_75":51,"_76":-5,"_99":213,"_18":214},{"_36":201,"_13":202,"_62":203,"_64":194,"_66":67,"_68":34,"_69":70,"_38":204,"_17":32,"_72":97,"_74":205,"_75":51,"_76":-5,"_99":206,"_18":207},"0cfd2b9e-0a11-11f1-8914-42dde27851f2","schema.ts","019c5f00-3df9-72b9-8e7f-0d1c7ee17a28",["D",1771120376884],"/**\n * Drizzle ORM Schema for Claude Plugins API\n */\nimport { sql } from \"npm:drizzle-orm@0.44.2\";\nimport { integer, sqliteTable, text } from \"npm:drizzle-orm@0.44.2/sqlite-core\";\n\n/**\n * Plugins table - stores all registered plugins\n */\nexport const plugins = sqliteTable(\"plugins\", {\n id: text(\"id\").primaryKey(),\n name: text(\"name\").notNull(),\n namespace: text(\"namespace\").notNull(),\n gitUrl: text(\"git_url\").notNull(),\n\n // Search fields\n description: text(\"description\"),\n keywords: text(\"keywords\"), // JSON stringified array\n category: text(\"category\"),\n skills: text(\"skills\"), // JSON stringified array\n\n // Display metadata\n version: text(\"version\").notNull().default(\"1.0.0\"),\n author: text(\"author\").notNull(), // Simple string, no complex object\n\n // All other marketplace data as single JSON blob\n metadata: text(\"metadata\"), // JSON: { homepage, license, commands, agents, mcpServers, etc. }\n\n // GitHub signals\n stars: integer(\"stars\").notNull().default(0),\n\n // Plugin status\n verified: integer(\"verified\", { mode: \"boolean\" }).notNull().default(false),\n downloads: integer(\"downloads\").notNull().default(0),\n\n // Timestamps\n createdAt: text(\"created_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n updatedAt: text(\"updated_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n});\n\n/**\n * Plugin stats table - tracks download statistics\n */\nexport const pluginStats = sqliteTable(\"plugin_stats\", {\n pluginId: text(\"plugin_id\")\n .primaryKey()\n .references(() => plugins.id),\n downloadsTotal: integer(\"downloads_total\").notNull().default(0),\n downloadsWeek: integer(\"downloads_week\").notNull().default(0),\n downloadsMonth: integer(\"downloads_month\").notNull().default(0),\n lastDownloadedAt: text(\"last_downloaded_at\"),\n});\n\n/**\n * Search Buckets tabs - stores all search buckets\n */\nexport const searchBuckets = sqliteTable(\"search_buckets\", {\n id: text(\"id\").primaryKey(),\n name: text(\"name\").notNull(),\n buckets: text(\"buckets\").notNull(), // JSON stringified array\n\n // Timestamps\n createdAt: text(\"created_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n updatedAt: text(\"updated_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n});\n\n/**\n * Skills table - stores all registered Claude Code Skills\n */\nexport const skills = sqliteTable(\"skills\", {\n id: text(\"id\").primaryKey(),\n name: text(\"name\").notNull(),\n namespace: text(\"namespace\").notNull(), // @owner/repo/skillName\n sourceUrl: text(\"source_url\").notNull(), // GitHub tree URL\n sourceType: text(\"source_type\").notNull().default(\"github\"), // \"github\" | \"r2\"\n\n // From SKILL.md YAML frontmatter\n description: text(\"description\").notNull(), // Max 200 chars\n version: text(\"version\"),\n dependencies: text(\"dependencies\"),\n\n // Metadata\n author: text(\"author\").notNull(),\n stars: integer(\"stars\").notNull().default(0),\n installs: integer(\"installs\").notNull().default(0),\n verified: integer(\"verified\", { mode: \"boolean\" }).notNull().default(false),\n deleted: integer(\"deleted\", { mode: \"boolean\" }).notNull().default(false),\n duplicate: integer(\"duplicate\", { mode: \"boolean\" }).notNull().default(false),\n\n // JSON blob for extra data\n metadata: text(\"metadata\"), // {repoOwner, repoName, directoryPath, rawFileUrl, license, ...}\n\n // Timestamps\n createdAt: text(\"created_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n updatedAt: text(\"updated_at\").notNull().default(sql`CURRENT_TIMESTAMP`),\n firstCommitAt: text(\"first_commit_at\"), // When SKILL.md was first committed (for deduplication)\n contentHash: text(\"content_hash\"), // SHA-256 hash of SKILL.md content (for duplicate detection)\n lastCheckedAt: text(\"last_checked_at\"), // Last time health check ran on this skill\n});\n\n/**\n * Skill stats table - tracks install statistics\n */\nexport const skillStats = sqliteTable(\"skill_stats\", {\n skillId: text(\"skill_id\")\n .primaryKey()\n .references(() => skills.id),\n installsTotal: integer(\"installs_total\").notNull().default(0),\n installsWeek: integer(\"installs_week\").notNull().default(0),\n installsMonth: integer(\"installs_month\").notNull().default(0),\n lastInstalledAt: text(\"last_installed_at\"),\n});\n\nexport const skillsFts = sqliteTable(\"skills_fts\", {\n skillId: text(\"skill_id\"),\n name: text(\"name\"),\n description: text(\"description\"),\n namespace: text(\"namespace\"),\n author: text(\"author\"),\n});\n\nexport const pluginsFts = sqliteTable(\"plugins_fts\", {\n pluginId: text(\"plugin_id\"),\n name: text(\"name\"),\n description: text(\"description\"),\n keywords: text(\"keywords\"),\n namespace: text(\"namespace\"),\n skills: text(\"skills\"),\n});\n\n// Type exports for TypeScript\nexport type Plugin = typeof plugins.$inferSelect;\nexport type NewPlugin = typeof plugins.$inferInsert;\n\nexport type PluginStats = typeof pluginStats.$inferSelect;\nexport type NewPluginStats = typeof pluginStats.$inferInsert;\n\nexport type Skill = typeof skills.$inferSelect;\nexport type NewSkill = typeof skills.$inferInsert;\n\nexport type SkillStats = typeof skillStats.$inferSelect;\nexport type NewSkillStats = typeof skillStats.$inferInsert;\n\nexport type SkillsFts = typeof skillsFts.$inferSelect;\nexport type PluginsFts = typeof pluginsFts.$inferSelect;",[],[193,202],"0cfd28c4-0a11-11f1-8914-42dde27851f2","run-migrations.ts","019c5f00-3df9-72b9-8e7f-08206689825e",["D",1771120376884],"import { runMigrations } from \"./db.ts\";\n\n(async function run() {\n try {\n await runMigrations();\n console.log(\"Migrations completed successfully\");\n } catch (error) {\n console.error(\"Error running migrations:\", error);\n throw error;\n }\n})();",[],[193,209],"0cfcd040-0a11-11f1-8914-42dde27851f2","db.ts","019c5f00-3df9-72b9-8e7f-070531445f9f",["D",1771120376884],"/**\n * Database connection and utilities\n */\nimport { sqlite } from \"../clients/sqlite.ts\";\nimport { sql } from \"npm:drizzle-orm@0.44.2\";\nimport { drizzle } from \"npm:drizzle-orm@0.44.2/libsql\";\nimport * as schema from \"./schema.ts\";\n\n// Initialize Drizzle with Val Town SQLite (without schema to avoid circular refs)\nexport const db = drizzle(sqlite as any);\n\nasync function runCreateSearchBuckets() {\n console.log(\"Running create search buckets\");\n try {\n // Check if tables exist by attempting to query them\n const tableCheck = await db\n .select({ count: sql`count(*)` })\n .from(schema.searchBuckets)\n .limit(1)\n .catch(() => null);\n\n if (tableCheck === null) {\n console.log(\"Creating table for Search Buckets...\");\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS search_buckets (\n id TEXT PRIMARY KEY,\n name TEXT NOT NULL,\n buckets TEXT,\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )\n `);\n console.log(\"Search buckets created successfully\");\n }\n } catch (e) {\n console.error(\"Table creation failed\");\n throw e;\n }\n}\n\n/**\n * Run database migrations to create tables and indexes\n */\nasync function runCreateTablesAndIndexes() {\n console.log(\"Running create tables and indexes...\");\n\n try {\n // Check if tables exist by attempting to query them\n const tableCheck = await db\n .select({ count: sql`count(*)` })\n .from(schema.plugins)\n .limit(1)\n .catch(() => null);\n\n if (tableCheck === null) {\n // Tables don't exist, create them using schema-driven SQL\n console.log(\"Creating database schema...\");\n\n // Create tables using Drizzle schema definitions\n // This ensures schema and database stay in sync\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS plugins (\n id TEXT PRIMARY KEY,\n name TEXT NOT NULL,\n namespace TEXT NOT NULL,\n git_url TEXT NOT NULL,\n description TEXT,\n keywords TEXT,\n category TEXT,\n version TEXT NOT NULL DEFAULT '1.0.0',\n author TEXT NOT NULL,\n metadata TEXT,\n stars INTEGER NOT NULL DEFAULT 0,\n verified INTEGER NOT NULL DEFAULT 0,\n downloads INTEGER NOT NULL DEFAULT 0,\n first_commit_at TEXT,\n content_hash TEXT,\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )\n `);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS plugin_stats (\n plugin_id TEXT PRIMARY KEY,\n downloads_total INTEGER NOT NULL DEFAULT 0,\n downloads_week INTEGER NOT NULL DEFAULT 0,\n downloads_month INTEGER NOT NULL DEFAULT 0,\n last_downloaded_at TEXT,\n FOREIGN KEY (plugin_id) REFERENCES plugins(id) ON DELETE CASCADE\n )\n `);\n\n // Create indexes for query performance\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_plugins_namespace ON plugins(namespace)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_plugins_name ON plugins(name)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_plugins_category ON plugins(category)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_plugins_content_hash ON plugins(content_hash)`,\n );\n await sqlite.execute(\n `CREATE UNIQUE INDEX IF NOT EXISTS idx_plugins_namespace_name ON plugins(namespace, name)`,\n );\n\n console.log(\"Schema created successfully\");\n }\n\n console.log(\"Creating database completed successfully\");\n } catch (error) {\n console.error(\"Migration failed:\", error);\n throw error;\n }\n}\n\nasync function runAddSkillsMigration(): Promise {\n console.log(\"Running migration: add `skills` column to `plugins`…\");\n const pragma = await sqlite.execute(`PRAGMA table_info(plugins);`);\n const hasSkills = Array.isArray(pragma?.rows)\n ? pragma.rows.some((r: any) => r.name === \"skills\")\n : false;\n\n if (hasSkills) {\n console.log(\"Column `skills` already exists. Skipping ALTER TABLE.\");\n } else {\n console.log(\"Adding `skills` column (TEXT, nullable) to `plugins`…\");\n await sqlite.execute(`ALTER TABLE plugins ADD COLUMN skills TEXT;`);\n console.log(\"Column `skills` added.\");\n }\n\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_plugins_skills ON plugins(skills);`,\n );\n\n console.log(\"Migration completed: add `skills` column.\");\n}\n\nasync function runAddSkillsTableMigration(): Promise {\n console.log(\"Running migration: create skills and skill_stats tables...\");\n\n try {\n // Check if skills table exists\n const skillsTableCheck = await sqlite\n .execute(\n `SELECT name FROM sqlite_master WHERE type='table' AND name='skills';`,\n )\n .catch(() => null);\n\n if (!skillsTableCheck || skillsTableCheck.rows.length === 0) {\n console.log(\"Creating skills table...\");\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS skills (\n id TEXT PRIMARY KEY,\n name TEXT NOT NULL,\n namespace TEXT NOT NULL,\n source_url TEXT NOT NULL,\n source_type TEXT NOT NULL DEFAULT 'github',\n description TEXT NOT NULL,\n version TEXT,\n dependencies TEXT,\n author TEXT NOT NULL,\n stars INTEGER NOT NULL DEFAULT 0,\n installs INTEGER NOT NULL DEFAULT 0,\n verified INTEGER NOT NULL DEFAULT 0,\n metadata TEXT,\n first_commit_at TEXT,\n content_hash TEXT,\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )\n `);\n\n await sqlite.execute(`\n CREATE TABLE IF NOT EXISTS skill_stats (\n skill_id TEXT PRIMARY KEY,\n installs_total INTEGER NOT NULL DEFAULT 0,\n installs_week INTEGER NOT NULL DEFAULT 0,\n installs_month INTEGER NOT NULL DEFAULT 0,\n last_installed_at TEXT,\n FOREIGN KEY (skill_id) REFERENCES skills(id) ON DELETE CASCADE\n )\n `);\n\n // Create indexes\n await sqlite.execute(\n `CREATE UNIQUE INDEX IF NOT EXISTS idx_skills_namespace ON skills(namespace)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_skills_name ON skills(name)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_skills_author ON skills(author)`,\n );\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_skills_content_hash ON skills(content_hash)`,\n );\n\n console.log(\"Skills tables created successfully\");\n } else {\n console.log(\"Skills table already exists. Skipping.\");\n }\n } catch (error) {\n console.error(\"Skills table migration failed:\", error);\n throw error;\n }\n\n console.log(\"Migration completed: skills tables\");\n}\n\nasync function runAddDeletedColumnMigration(): Promise {\n console.log(\"Running migration: add `deleted` and `deleted_at` to skills...\");\n try {\n const skillsTableCheck = await sqlite\n .execute(\n `SELECT name FROM sqlite_master WHERE type='table' AND name='skills';`,\n )\n .catch(() => null);\n\n if (!skillsTableCheck) {\n throw new Error(\"Run runAddSkillsTableMigration before this migration\");\n }\n\n const pragma = await sqlite.execute(`PRAGMA table_info(skills);`);\n const cols = new Set(\n (pragma?.rows ?? []).map((r: any) =>\n (r?.name ?? r?.[1] ?? \"\").toLowerCase()\n ),\n );\n if (!cols.has(\"deleted\")) {\n console.log(\"Adding `deleted` column to skills...\");\n await sqlite.execute(\n `ALTER TABLE skills ADD COLUMN deleted INTEGER NOT NULL DEFAULT 0;`,\n );\n } else {\n console.log(\"`deleted` column already exists. Skipping.\");\n }\n } catch (e) {\n console.error(\"Skills table migration failed:\", e);\n throw e;\n }\n}\n\n/**\n * FTS5 Full-Text Search migration for skills table\n * \n * Uses CONTENTLESS FTS5 (no content='...' option) to avoid:\n * - Turso/libSQL auto-populate bug with external content tables\n * - Rowid sync issues between FTS and skills table\n * \n * Includes skill_id column for joining back to skills table.\n * Triggers are scoped to only fire on text column changes.\n */\nasync function runCreateSkillsFTS5Migration(): Promise {\n console.log(\"Running migration: create FTS5 virtual table for skills...\");\n\n try {\n // Check if FTS table already exists\n const ftsTableCheck = await sqlite\n .execute(\n `SELECT name FROM sqlite_master WHERE type='table' AND name='skills_fts';`,\n )\n .catch(() => null);\n\n if (ftsTableCheck && ftsTableCheck.rows.length > 0) {\n console.log(\"skills_fts table already exists. Skipping creation.\");\n return;\n }\n\n // Create CONTENTLESS FTS5 virtual table\n // No content='...' option - FTS stores its own copy of text\n // Includes skill_id for joining back to skills table\n console.log(\"Creating FTS5 virtual table (contentless)...\");\n await sqlite.execute(`\n CREATE VIRTUAL TABLE IF NOT EXISTS skills_fts USING fts5(\n skill_id,\n name,\n description,\n namespace,\n author,\n tokenize='porter unicode61 remove_diacritics 1'\n );\n `);\n\n // Populate FTS table from existing skills\n console.log(\"Populating FTS5 table with existing skills...\");\n await sqlite.execute(`\n INSERT INTO skills_fts(skill_id, name, description, namespace, author)\n SELECT id, name, description, namespace, author\n FROM skills\n WHERE deleted = 0 AND duplicate = 0;\n `);\n\n // Create triggers to keep FTS in sync\n console.log(\"Creating triggers for FTS5 sync...\");\n\n // After INSERT trigger\n await sqlite.execute(`\n CREATE TRIGGER IF NOT EXISTS skills_ai AFTER INSERT ON skills\n WHEN NEW.deleted = 0 AND NEW.duplicate = 0\n BEGIN\n INSERT INTO skills_fts(skill_id, name, description, namespace, author)\n VALUES (NEW.id, NEW.name, NEW.description, NEW.namespace, NEW.author);\n END;\n `);\n\n // After DELETE trigger - only for rows that were in FTS\n await sqlite.execute(`\n CREATE TRIGGER IF NOT EXISTS skills_ad AFTER DELETE ON skills\n WHEN OLD.deleted = 0 AND OLD.duplicate = 0\n BEGIN\n DELETE FROM skills_fts WHERE skill_id = OLD.id;\n END;\n `);\n\n // After UPDATE trigger - only fires when FTS-relevant columns change\n // This prevents install increments and other non-text updates from touching FTS\n await sqlite.execute(`\n CREATE TRIGGER IF NOT EXISTS skills_au AFTER UPDATE ON skills\n WHEN OLD.name != NEW.name \n OR OLD.description != NEW.description \n OR OLD.namespace != NEW.namespace \n OR OLD.author != NEW.author\n OR OLD.deleted != NEW.deleted\n OR OLD.duplicate != NEW.duplicate\n BEGIN\n DELETE FROM skills_fts WHERE skill_id = OLD.id;\n INSERT INTO skills_fts(skill_id, name, description, namespace, author)\n SELECT NEW.id, NEW.name, NEW.description, NEW.namespace, NEW.author\n WHERE NEW.deleted = 0 AND NEW.duplicate = 0;\n END;\n `);\n\n console.log(\"FTS5 migration completed successfully\");\n } catch (error) {\n console.error(\"FTS5 migration failed:\", error);\n throw error;\n }\n}\n\n/**\n * Rebuild FTS5 index from scratch\n * Use this if FTS gets out of sync with skills table\n */\nexport async function rebuildSkillsFTS(): Promise {\n console.log(\"Rebuilding skills_fts index...\");\n\n // Delete all existing FTS data\n await sqlite.execute(`DELETE FROM skills_fts;`);\n\n // Repopulate from skills table using skill_id\n await sqlite.execute(`\n INSERT INTO skills_fts(skill_id, name, description, namespace, author)\n SELECT id, name, description, namespace, author\n FROM skills\n WHERE deleted = 0 AND duplicate = 0;\n `);\n\n console.log(\"FTS5 index rebuilt successfully\");\n}\n\n\n/**\n * Add lastCheckedAt column to skills table for health check tracking\n */\nasync function runAddLastCheckedAtMigration(): Promise {\n console.log(\"Running migration: add `last_checked_at` to skills...\");\n try {\n const pragma = await sqlite.execute(`PRAGMA table_info(skills);`);\n const cols = new Set(\n (pragma?.rows ?? []).map((r: any) =>\n (r?.name ?? r?.[1] ?? \"\").toLowerCase()\n ),\n );\n\n if (!cols.has(\"last_checked_at\")) {\n console.log(\"Adding `last_checked_at` column to skills...\");\n await sqlite.execute(\n `ALTER TABLE skills ADD COLUMN last_checked_at TEXT;`,\n );\n\n // Create index for efficient querying by lastCheckedAt\n await sqlite.execute(\n `CREATE INDEX IF NOT EXISTS idx_skills_last_checked_at ON skills(last_checked_at);`,\n );\n\n console.log(\"`last_checked_at` column added successfully.\");\n } else {\n console.log(\"`last_checked_at` column already exists. Skipping.\");\n }\n } catch (e) {\n console.error(\"last_checked_at migration failed:\", e);\n throw e;\n }\n}\n\nasync function runAddpartialIndexesMigration() {\n console.log('Running migration: add partial indxes for query optimization')\n\n try {\n await sqlite.execute(`\n CREATE INDEX IF NOT EXISTS idx_skills_health_queue\n ON skills(last_checked_at)\n WHERE deleted = 0 AND duplicate = 0;\n `)\n console.log(\"Created idx_skills_health_queue\");\n\n // Partial index for active skills listing (ordered by installs, stars)\n // This eliminates full table scans for search/listing without query\n await sqlite.execute(`\n CREATE INDEX IF NOT EXISTS idx_skills_active\n ON skills(installs DESC, stars DESC, id)\n WHERE deleted = 0 AND duplicate = 0;\n `)\n\n console.log(\"Created idx_skills_active_order\");\n // Composite index for namespace lookups (already unique but ensure it exists)\n await sqlite.execute(\n `CREATE UNIQUE INDEX IF NOT EXISTS idx_skills_namespace ON skills(namespace);`\n );\n\n console.log(\"Partial indexes migration completed successfully\"); \n } catch (error) {\n console.error('Partial indexes migration failed', error)\n }\n}\n\nasync function runPluginsFTSAndCoveringIndexesMigration(): Promise {\n console.log(\"Running migration: plugins FTS5 and covering indexes...\");\n\n try {\n // 1. Create covering indexes (These were fine!)\n console.log(\"Creating covering indexes...\");\n await sqlite.execute(`\n CREATE INDEX IF NOT EXISTS idx_skills_active_covering \n ON skills(installs DESC, stars DESC)\n WHERE deleted = 0 AND duplicate = 0;\n `);\n\n await sqlite.execute(`\n CREATE INDEX IF NOT EXISTS idx_plugins_downloads_stars \n ON plugins(downloads DESC, stars DESC);\n `);\n\n // 2. Cleanup: Drop old table/triggers if they exist to ensure clean state\n await sqlite.execute(`DROP TABLE IF EXISTS plugins_fts;`);\n await sqlite.execute(`DROP TRIGGER IF EXISTS plugins_ai;`);\n await sqlite.execute(`DROP TRIGGER IF EXISTS plugins_ad;`);\n await sqlite.execute(`DROP TRIGGER IF EXISTS plugins_au;`);\n\n // 3. Create FTS5 virtual table (Contentless)\n // CHANGE: Removed 'plugin_id'. We will use the built-in 'rowid' instead.\n console.log(\"Creating plugins_fts FTS5 virtual table...\");\n await sqlite.execute(`\n CREATE VIRTUAL TABLE plugins_fts USING fts5(\n name,\n description,\n keywords,\n namespace,\n skills,\n tokenize='porter unicode61 remove_diacritics 1',\n content='' -- Contentless mode\n );\n `);\n\n // 4. Populate FTS using the Main Table's rowid\n // CHANGE: Mapping 'rowid' explicitly ensures O(1) lookups later\n console.log(\"Populating plugins_fts...\");\n await sqlite.execute(`\n INSERT INTO plugins_fts(rowid, name, description, keywords, namespace, skills)\n SELECT rowid, name, description, keywords, namespace, skills\n FROM plugins;\n `);\n\n // 5. Create Triggers (The Complex Part)\n console.log(\"Creating triggers for plugins_fts sync...\");\n\n // INSERT: Sync the rowid\n await sqlite.execute(`\n CREATE TRIGGER plugins_ai AFTER INSERT ON plugins\n BEGIN\n INSERT INTO plugins_fts(rowid, name, description, keywords, namespace, skills)\n VALUES (NEW.rowid, NEW.name, NEW.description, NEW.keywords, NEW.namespace, NEW.skills);\n END;\n `);\n\n // DELETE: Use 'delete' magic command (Required for Contentless tables)\n await sqlite.execute(`\n CREATE TRIGGER plugins_ad AFTER DELETE ON plugins\n BEGIN\n INSERT INTO plugins_fts(plugins_fts, rowid, name, description, keywords, namespace, skills)\n VALUES('delete', OLD.rowid, OLD.name, OLD.description, OLD.keywords, OLD.namespace, OLD.skills);\n END;\n `);\n\n // UPDATE: Delete old entries + Insert new ones\n await sqlite.execute(`\n CREATE TRIGGER plugins_au AFTER UPDATE ON plugins\n WHEN OLD.name != NEW.name \n OR OLD.description != NEW.description \n OR OLD.keywords != NEW.keywords\n OR OLD.namespace != NEW.namespace\n OR OLD.skills != NEW.skills\n BEGIN\n -- Step 1: Delete old (Magic Syntax)\n INSERT INTO plugins_fts(plugins_fts, rowid, name, description, keywords, namespace, skills)\n VALUES('delete', OLD.rowid, OLD.name, OLD.description, OLD.keywords, OLD.namespace, OLD.skills);\n \n -- Step 2: Insert new\n INSERT INTO plugins_fts(rowid, name, description, keywords, namespace, skills)\n VALUES (NEW.rowid, NEW.name, NEW.description, NEW.keywords, NEW.namespace, NEW.skills);\n END;\n `);\n\n console.log(\"Plugins FTS5 migration completed successfully\");\n } catch (error) {\n console.error(\"Plugins FTS5 migration failed:\", error);\n throw error;\n }\n}\n\nasync function fixSkillsFTSMigration(): Promise {\n console.log(\"Running migration: REPAIR skills_fts...\");\n\n try {\n // 1. Drop the broken setup (Triggers first!)\n await sqlite.execute(`DROP TRIGGER IF EXISTS skills_ai;`);\n await sqlite.execute(`DROP TRIGGER IF EXISTS skills_ad;`);\n await sqlite.execute(`DROP TRIGGER IF EXISTS skills_au;`);\n await sqlite.execute(`DROP TABLE IF EXISTS skills_fts;`);\n\n // 2. Recreate correctly (No skill_id column, use rowid, contentless)\n await sqlite.execute(`\n CREATE VIRTUAL TABLE skills_fts USING fts5(\n name,\n description,\n namespace,\n author,\n tokenize='porter unicode61 remove_diacritics 1',\n content='' \n );\n `);\n\n // 3. Repopulate (using rowid mapping)\n console.log(\"Repopulating skills_fts...\");\n await sqlite.execute(`\n INSERT INTO skills_fts(rowid, name, description, namespace, author)\n SELECT rowid, name, description, namespace, author\n FROM skills\n WHERE deleted = 0 AND duplicate = 0;\n `);\n\n // 4. Recreate Triggers (using rowid and 'delete' magic)\n \n // INSERT\n await sqlite.execute(`\n CREATE TRIGGER skills_ai AFTER INSERT ON skills\n WHEN NEW.deleted = 0 AND NEW.duplicate = 0\n BEGIN\n INSERT INTO skills_fts(rowid, name, description, namespace, author)\n VALUES (NEW.rowid, NEW.name, NEW.description, NEW.namespace, NEW.author);\n END;\n `);\n\n // DELETE\n await sqlite.execute(`\n CREATE TRIGGER skills_ad AFTER DELETE ON skills\n WHEN OLD.deleted = 0 AND OLD.duplicate = 0\n BEGIN\n INSERT INTO skills_fts(skills_fts, rowid, name, description, namespace, author)\n VALUES('delete', OLD.rowid, OLD.name, OLD.description, OLD.namespace, OLD.author);\n END;\n `);\n\n // UPDATE (The full logic)\n await sqlite.execute(`\n CREATE TRIGGER skills_au AFTER UPDATE ON skills\n WHEN OLD.name != NEW.name \n OR OLD.description != NEW.description \n OR OLD.namespace != NEW.namespace\n OR OLD.author != NEW.author\n OR OLD.deleted != NEW.deleted\n OR OLD.duplicate != NEW.duplicate\n BEGIN\n -- 1. Remove old entry (Magic 'delete')\n INSERT INTO skills_fts(skills_fts, rowid, name, description, namespace, author)\n VALUES('delete', OLD.rowid, OLD.name, OLD.description, OLD.namespace, OLD.author);\n \n -- 2. Add new entry (if not deleted/duplicate)\n INSERT INTO skills_fts(rowid, name, description, namespace, author)\n SELECT NEW.rowid, NEW.name, NEW.description, NEW.namespace, NEW.author\n WHERE NEW.deleted = 0 AND NEW.duplicate = 0;\n END;\n `);\n\n console.log(\"REPAIR of skills_fts completed.\");\n } catch (error) {\n console.error(\"Skills Repair failed:\", error);\n throw error;\n }\n}\n\nexport async function runMigrations(): Promise {\n // await runCreateTablesAndIndexes();\n // await runAddSkillsMigration();\n // await runAddSkillsTableMigration();\n // await runCreateSearchBuckets()\n // await runAddDeletedColumnMigration();\n // await runCreateSkillsFTS5Migration();\n // await runAddLastCheckedAtMigration();\n // await runAddpartialIndexesMigration();\n \n await runPluginsFTSAndCoveringIndexesMigration();\n await fixSkillsFTSMigration();\n}",[],[193,216],"0cfcced8-0a11-11f1-8914-42dde27851f2","cron","019c5f00-3df9-72b9-8e7f-00a373d269d7",["D",1771120376884],[228,229,230,231,232,233],[223],{"_36":270,"_13":271,"_62":272,"_64":224,"_66":67,"_68":34,"_69":70,"_38":273,"_17":32,"_72":97,"_74":274,"_75":51,"_76":-5,"_99":275,"_18":276},{"_36":263,"_13":264,"_62":265,"_64":224,"_66":67,"_68":34,"_69":70,"_38":266,"_17":32,"_72":97,"_74":267,"_75":51,"_76":-5,"_99":268,"_18":269},{"_36":256,"_13":257,"_62":258,"_64":224,"_66":67,"_68":34,"_69":70,"_38":259,"_17":32,"_72":238,"_74":260,"_75":51,"_76":-5,"_99":261,"_18":262},{"_36":249,"_13":250,"_62":251,"_64":224,"_66":67,"_68":34,"_69":70,"_38":252,"_17":32,"_72":238,"_74":253,"_75":51,"_76":-5,"_99":254,"_18":255},{"_36":242,"_13":243,"_62":244,"_64":224,"_66":67,"_68":34,"_69":70,"_38":245,"_17":32,"_72":238,"_74":246,"_75":51,"_76":-5,"_99":247,"_18":248},{"_36":234,"_13":235,"_62":236,"_64":224,"_66":67,"_68":34,"_69":70,"_38":237,"_17":32,"_72":238,"_74":239,"_75":51,"_76":-5,"_99":240,"_18":241},"0cfcaf8e-0a11-11f1-8914-42dde27851f2","skills-indexer.ts","019c5f00-3df9-72b9-8e7e-fc10b8abc1a5",["D",1771120376884],"interval","import {\n Bucket,\n GitHubClient,\n GitHubSearchResponse,\n} from \"../clients/github.ts\";\nimport { blob } from \"https://esm.town/v/std/blob\";\nimport {\n Logger,\n withLogger,\n} from \"https://esm.town/v/kamalnrf/axiom-logger/standalone.ts\";\nimport { indexSkillFiles } from \"../lib/skills/index-skill-files.ts\";\nimport type { SkillFileRef } from \"../lib/skills/types.ts\";\n\ntype SkillsIndexerRunState = {\n lastIndexProcessed: number; // exclusive end of last batch (next run starts here)\n batchSize: number;\n totalBuckets: number;\n sortedBy: \"count_desc_min_asc\";\n startedAt: string;\n completedAt?: string;\n};\nconst LAST_RUN_STATE_KEY = \"skills-indexer:last-run-state\";\nexport async function getLastRunState(): Promise {\n try {\n return await blob.getJSON(LAST_RUN_STATE_KEY);\n } catch {\n return null;\n }\n}\n\nfunction computeEndIndexByItemCount(\n buckets: Array,\n start: number,\n maxItems: number,\n): number {\n let total = 0;\n let i = start;\n while (i < buckets.length && total + buckets[i].count <= maxItems) {\n total += buckets[i].count;\n i++;\n }\n // If the very first bucket exceeds maxItems, still process at least one bucket\n return i === start ? Math.min(start + 1, buckets.length) : i;\n}\n\nexport async function setLastRunState(\n state: SkillsIndexerRunState,\n): Promise {\n await blob.setJSON(LAST_RUN_STATE_KEY, state);\n}\n\nconst github = new GitHubClient();\n\ntype IndexSkillsArgs = {\n minSize: number;\n maxSize: number;\n};\n\n/**\n * Search GitHub for SKILL.md files matching size criteria,\n * then delegate to shared indexSkillFiles pipeline\n */\nasync function indexSkills({ minSize, maxSize }: IndexSkillsArgs, log: Logger) {\n let pageSearchFailures = 0;\n\n // Search for SKILL.md files in skills/ directories (exclude archived repos)\n const searchQuery =\n `filename:SKILL.md language:Markdown \"---\" \"name:\" \"description:\" size:${minSize}..${maxSize}`;\n log.info(`Searching: ${searchQuery}`);\n\n const firstPage = await github.searchCode(searchQuery);\n log.info(`Found ${firstPage.total_count} SKILL.md files`);\n\n // Calculate total pages (max 10 pages = 1000 results)\n const maxPages = Math.min(Math.ceil(firstPage.total_count / 100), 10);\n if (firstPage.total_count > 1000) {\n log.warn(\n `Found ${firstPage.total_count} results. GitHub limits to 1000, only first 1000 will be indexed.`,\n );\n }\n\n // Collect all skill file references from search results\n const allRefs: SkillFileRef[] = [];\n\n for (let page = 1; page <= maxPages; page++) {\n log.info(`\\nProcessing page ${page}/${maxPages}...`);\n\n let searchResults: GitHubSearchResponse;\n try {\n searchResults = page === 1\n ? firstPage\n : await github.searchCode(searchQuery, page);\n } catch (error) {\n console.error(\n `Failed to fetch search results for page ${page}:`,\n error,\n );\n pageSearchFailures++;\n continue;\n }\n\n log.info(`Found ${searchResults.items.length} items on page ${page}`);\n\n // Transform search results to SkillFileRef\n for (const item of searchResults.items) {\n allRefs.push({\n repoFullName: item.repository.full_name,\n repoHtmlUrl: item.repository.html_url,\n path: item.path,\n });\n }\n }\n\n if (allRefs.length === 0) {\n log.info(\"No skills found in search\");\n return {\n indexed: 0,\n invalid: 0,\n failed: pageSearchFailures,\n };\n }\n\n log.info(`Collected ${allRefs.length} skill file references from search`);\n\n // Delegate to shared indexing pipeline\n const result = await indexSkillFiles(allRefs, { github, log });\n\n return {\n indexed: result.indexed,\n invalid: result.invalid,\n failed: result.failed + pageSearchFailures,\n };\n}\n\nexport default withLogger({\n valName: \"skills-indexer\",\n console: \"none\",\n type: \"CRON\",\n}, async (log) => {\n const startTime = Date.now();\n\n try {\n log.info(\"GitHub Skills indexer started\");\n const skillsSearchBuckets = await blob.getJSON(\n \"skills-search-buckets\",\n );\n log.info(\"skillSearchBuckets =>\", { skillsSearchBuckets });\n const sortedBuckets = skillsSearchBuckets.sort((a, b) => {\n if (b.count !== a.count) return b.count - a.count;\n return a.min - b.min;\n });\n\n const BATCH_SIZE = 4000;\n const prevState = await getLastRunState();\n const startIdx = Math.min(\n Math.max(prevState?.lastIndexProcessed ?? 0, 0),\n sortedBuckets.length,\n );\n const endIdx = computeEndIndexByItemCount(\n sortedBuckets,\n startIdx,\n BATCH_SIZE,\n );\n\n const runState: SkillsIndexerRunState = {\n lastIndexProcessed: endIdx,\n batchSize: BATCH_SIZE,\n totalBuckets: sortedBuckets.length,\n sortedBy: \"count_desc_min_asc\",\n startedAt: (new Date(startTime)).toISOString(),\n };\n console.log(\"Run State =>\", runState);\n await setLastRunState(runState);\n const currentBatch = sortedBuckets.slice(startIdx, endIdx);\n console.log(\n `Processing buckets ${startIdx}..${endIdx} of ${sortedBuckets.length} (batch size ${currentBatch.length})`,\n );\n log.info(\n `Processing buckets ${startIdx}..${endIdx} of ${sortedBuckets.length} (batch size ${currentBatch.length})`,\n );\n\n let indexed = 0;\n let invalid = 0;\n let failed = 0;\n\n const results = await Promise.all(\n currentBatch.map(async ({ min, max }) => {\n try {\n return await indexSkills({ minSize: min, maxSize: max }, log);\n } catch (e) {\n console.error(\"indexSkills failed:\", e);\n return { indexed: 0, invalid: 0, failed: 1 };\n }\n }),\n );\n\n for (const res of results) {\n indexed += res.indexed;\n invalid += res.invalid;\n failed += res.failed;\n }\n\n const duration = ((Date.now() - startTime) / 1000).toFixed(2);\n console.log(\n `Skills indexer completed in ${duration}s: ✓ Indexed ${indexed} skills, ${invalid} invalid, ${failed} failed.`,\n );\n\n // Update Final Run State\n const completedState: SkillsIndexerRunState = {\n ...runState,\n completedAt: new Date().toISOString(),\n };\n\n if (endIdx >= sortedBuckets.length) {\n const resetState: SkillsIndexerRunState = {\n ...completedState,\n lastIndexProcessed: 0,\n };\n await setLastRunState(resetState);\n log.info(\"All buckets processed. Checkpoint reset to 0.\");\n } else {\n await setLastRunState(completedState);\n }\n } catch (e) {\n log.error(\"Skills indexer CRON job failed:\", { error: String(e) });\n throw e;\n }\n});",[],[223,235],"0cfc8a04-0a11-11f1-8914-42dde27851f2","skill-health.ts","019c5f00-3df9-72b9-8e7e-f8931bb8802e",["D",1771120376884],"/**\n * Unified Skill Health Check\n *\n * Runs:\n * 1. HEAD request to check if skill still exists\n * 2. If deleted → mark deleted, promote oldest duplicate\n * 3. If missing firstCommitAt → backfill it\n * 4. Update lastCheckedAt\n */\nimport { db } from \"../database/db.ts\";\nimport { skills } from \"../database/schema.ts\";\nimport {\n and,\n asc,\n eq,\n inArray,\n isNull,\n lt,\n or,\n sql,\n} from \"npm:drizzle-orm@0.44.2\";\nimport { blob } from \"https://esm.town/v/std/blob\";\nimport { chunkArray } from \"../utils.ts\";\nimport { GitHubClient } from \"../clients/github.ts\";\nimport {\n type Logger,\n withLogger,\n} from \"https://esm.town/v/kamalnrf/axiom-logger/standalone.ts\";\n\ntype RunState = {\n lastProcessedId: string | null;\n processedCount: number;\n deletedCount: number;\n promotedCount: number;\n backfilledCount: number;\n startedAt: string;\n completedAt?: string;\n};\n\nconst LAST_RUN_STATE_KEY = \"skill-health:last-run-state\";\nconst BATCH_SIZE = 10000;\nconst CHECK_INTERVAL_HOURS = 24; // Only re-check skills older than this\n\nasync function getLastRunState(): Promise {\n try {\n return await blob.getJSON(LAST_RUN_STATE_KEY);\n } catch {\n return null;\n }\n}\n\nasync function setLastRunState(state: RunState): Promise {\n await blob.setJSON(LAST_RUN_STATE_KEY, state);\n}\n\n/**\n * Check if a skill's raw file URL still exists\n */\nasync function checkSkillExists(rawFileUrl: string): Promise {\n try {\n const response = await fetch(rawFileUrl, { method: \"HEAD\" });\n return response.status !== 404;\n } catch {\n // Network error - assume still exists, will retry next run\n return true;\n }\n}\n\n/**\n * Mark a skill as deleted\n */\nasync function markSkillDeleted(skillId: string): Promise {\n await db\n .update(skills)\n .set({\n deleted: true,\n updatedAt: new Date().toISOString(),\n lastCheckedAt: new Date().toISOString(),\n })\n .where(eq(skills.id, skillId));\n}\n\n/**\n * When an original is deleted, find and promote the oldest duplicate with same contentHash\n * Returns true if a duplicate was promoted\n */\nasync function promoteOldestDuplicate(\n contentHash: string,\n deletedNamespace: string,\n log: Logger,\n): Promise {\n // Find duplicates with same content hash, ordered by firstCommitAt\n const duplicates = await db\n .select()\n .from(skills)\n .where(\n and(\n eq(skills.contentHash, contentHash),\n eq(skills.duplicate, true),\n eq(skills.deleted, false),\n ),\n )\n .orderBy(asc(skills.firstCommitAt))\n .limit(1);\n\n if (duplicates.length === 0) {\n log.info(`No duplicates to promote for ${deletedNamespace}`);\n return false;\n }\n\n const promoted = duplicates[0];\n\n // Promote this duplicate to original\n await db\n .update(skills)\n .set({\n duplicate: false,\n updatedAt: new Date().toISOString(),\n })\n .where(eq(skills.id, promoted.id));\n\n log.info(\n `Promoted ${promoted.namespace} as new original (was duplicate of ${deletedNamespace})`,\n );\n return true;\n}\n\n/**\n * Batch update lastCheckedAt for multiple skills\n * Reduces N individual UPDATEs to 1 batch UPDATE, saving significant row reads\n */\nasync function batchUpdateLastChecked(skillIds: string[]): Promise {\n if (skillIds.length === 0) return;\n\n await db\n .update(skills)\n .set({ lastCheckedAt: new Date().toISOString() })\n .where(inArray(skills.id, skillIds));\n}\n\n/**\n * Backfill firstCommitAt for a skill\n */\nasync function backfillFirstCommit(\n skill: typeof skills.$inferSelect,\n github: GitHubClient,\n log: Logger,\n): Promise {\n const metadata = JSON.parse(skill.metadata || \"{}\");\n\n if (!metadata.repoOwner || !metadata.repoName || !metadata.directoryPath) {\n return false;\n }\n\n const firstCommitAt = await github.fetchFirstCommitDate(\n `${metadata.repoOwner}/${metadata.repoName}`,\n `${metadata.directoryPath}/SKILL.md`,\n );\n\n if (firstCommitAt) {\n await db\n .update(skills)\n .set({ firstCommitAt })\n .where(eq(skills.id, skill.id));\n\n log.info(\n `Backfilled firstCommitAt for ${skill.namespace}: ${firstCommitAt}`,\n );\n return true;\n }\n\n return false;\n}\n\nexport default withLogger({\n valName: \"skill-health\",\n console: \"none\",\n type: \"CRON\",\n}, async (log) => {\n const startTime = Date.now();\n const startedAt = new Date().toISOString();\n\n let processedCount = 0;\n let deletedCount = 0;\n let promotedCount = 0;\n let backfilledCount = 0;\n\n try {\n log.info(\"Skill health check started\");\n\n const github = new GitHubClient();\n\n // Calculate cutoff time - only check skills not checked recently\n const cutoffTime = new Date(\n Date.now() - CHECK_INTERVAL_HOURS * 60 * 60 * 1000,\n ).toISOString();\n\n // Get originals (non-duplicate, non-deleted) that need checking\n // Either never checked, or checked more than CHECK_INTERVAL_HOURS ago\n const batch = await db\n .select()\n .from(skills)\n .where(\n and(\n eq(skills.deleted, false),\n eq(skills.duplicate, false),\n or(\n isNull(skills.lastCheckedAt),\n lt(skills.lastCheckedAt, cutoffTime),\n ),\n ),\n )\n .orderBy(asc(skills.lastCheckedAt)) // Oldest checked first\n .limit(BATCH_SIZE);\n\n if (batch.length === 0) {\n log.info(\"No skills need checking - all recently verified\");\n return;\n }\n\n log.info(`Processing ${batch.length} originals for health check`);\n\n for (const chunk of chunkArray(batch, 50)) {\n // Collect IDs to batch update lastCheckedAt\n const idsToMarkChecked: string[] = [];\n\n await Promise.allSettled(\n chunk.map(async (skill) => {\n try {\n const metadata = JSON.parse(skill.metadata || \"{}\");\n\n // 1. Check if skill still exists\n if (metadata.rawFileUrl) {\n const exists = await checkSkillExists(metadata.rawFileUrl);\n\n if (!exists) {\n log.info(`Skill deleted: ${skill.namespace}`);\n await markSkillDeleted(skill.id);\n deletedCount++;\n\n // Promote oldest duplicate if exists\n if (skill.contentHash) {\n const promoted = await promoteOldestDuplicate(\n skill.contentHash,\n skill.namespace,\n log,\n );\n if (promoted) promotedCount++;\n }\n\n processedCount++;\n return;\n }\n }\n\n // 2. Backfill firstCommitAt if missing\n if (!skill.firstCommitAt) {\n const backfilled = await backfillFirstCommit(skill, github, log);\n if (backfilled) backfilledCount++;\n }\n\n // 3. Collect for batch update (instead of individual UPDATE per skill)\n idsToMarkChecked.push(skill.id);\n processedCount++;\n } catch (error) {\n log.error(`Failed to process ${skill.namespace}:`, { error });\n }\n }),\n );\n\n // Batch update lastCheckedAt for all processed skills in this chunk\n await batchUpdateLastChecked(idsToMarkChecked);\n }\n\n const duration = ((Date.now() - startTime) / 1000).toFixed(2);\n\n const runState: RunState = {\n lastProcessedId: batch[batch.length - 1]?.id || null,\n processedCount,\n deletedCount,\n promotedCount,\n backfilledCount,\n startedAt,\n completedAt: new Date().toISOString(),\n };\n\n await setLastRunState(runState);\n\n log.info(`Skill health check completed in ${duration}s`, {\n processed: processedCount,\n deleted: deletedCount,\n promoted: promotedCount,\n backfilled: backfilledCount,\n });\n\n console.log(\n `Skill health: ${processedCount} checked, ${deletedCount} deleted, ${promotedCount} promoted, ${backfilledCount} backfilled in ${duration}s`,\n );\n } catch (error) {\n log.error(\"Skill health check failed:\", { error });\n throw error;\n }\n});",[],[223,243],"0cfc7cbc-0a11-11f1-8914-42dde27851f2","search-buckets.ts","019c5f00-3df9-72b9-8e7e-f5cdb718c7cd",["D",1771120376884],"import { GitHubClient } from \"../clients/github.ts\";\nimport { blob } from \"https://esm.town/v/std/blob\";\n\nconst MIN_SIZE = 500;\nconst MAX_SIZE = 150_000;\nconst BASE_QUERY =\n `filename:SKILL.md language:Markdown \"---\" \"name:\" \"description:\"`;\nconst BUCKET_NAME = \"skills-search-buckets\";\n\nconst MAX_PER_BUCKET = 950;\nconst MIN_SPAN = 200;\n\n// Helper to read previously stored buckets (returns [] if none)\nasync function getStoredBuckets(): Promise<\n Array<{ min: number; max: number; count: number }>\n> {\n try {\n const existing = await blob.getJSON<\n Array<{ min: number; max: number; count: number }>\n >(BUCKET_NAME);\n return Array.isArray(existing) ? existing : [];\n } catch {\n return [];\n }\n}\n\nexport default async function (interval: Interval) {\n const startTime = Date.now();\n\n try {\n console.log(\"GitHub Search Buckets for skills started\");\n const github = new GitHubClient();\n\n // 1) Load stored buckets. If empty, create initial buckets for full range.\n let buckets = await getStoredBuckets();\n if (buckets.length === 0) {\n console.log(\n \"No stored buckets found. Creating initial buckets from full size range.\",\n );\n buckets = await github.findBuckets(BASE_QUERY, {\n min: MIN_SIZE,\n max: MAX_SIZE,\n maxPerBucket: MAX_PER_BUCKET,\n minSpan: MIN_SPAN,\n });\n } else {\n console.log(\n `Loaded ${buckets.length} stored buckets. Refining oversized ones if needed…`,\n );\n }\n\n // Sort by Size Range\n buckets.sort((a, b) => (b.max - b.min) - (a.max - a.min));\n console.log(\"Existing Buckets =>\", buckets);\n\n // 2) Iterate stored (or newly created) buckets.\n // For any bucket with count > MAX_PER_BUCKET, reuse findBuckets over that bucket’s range to downsize.\n const refined: Array<{ min: number; max: number; count: number }> = [];\n for (const b of buckets) {\n const subs = await github.findBuckets(BASE_QUERY, {\n min: b.min,\n max: b.max,\n maxPerBucket: MAX_PER_BUCKET,\n minSpan: MIN_SPAN,\n });\n refined.push(...subs);\n }\n\n // 3) Persist the updated buckets\n await blob.setJSON(BUCKET_NAME, refined);\n\n const duration = ((Date.now() - startTime) / 1000).toFixed(2);\n console.log(\n `Search Buckets completed in ${duration}s, and created ${refined.length} buckets`,\n );\n } catch (e) {\n console.error(\"Skills indexer CRON job failed:\", e);\n throw e;\n }\n}",[],[223,250],"0cfc64d4-0a11-11f1-8914-42dde27851f2","plugins-indexer.ts","019c5f00-3df9-72b9-8e7e-f2d05e7b8b83",["D",1771120376884],"import { parseMarketplace, transformToDbPlugins } from \"./parser.ts\";\nimport { GitHubClient, GitHubSearchResponse } from \"../clients/github.ts\";\nimport { db } from \"../database/db.ts\";\nimport { plugins } from \"../database/schema.ts\";\nimport { sql } from \"npm:drizzle-orm@0.44.2\";\nimport { chunkArray } from \"../utils.ts\";\n\n// Learn more: https://docs.val.town/vals/cron/\nexport default async function (interval: Interval) {\n const startTime = Date.now();\n\n try {\n console.log(\"GitHub plugin indexer started\");\n const github = new GitHubClient();\n const searchQuery = \"filename:marketplace.json path:.claude-plugin\";\n\n const firstPage = await github.searchCode(searchQuery);\n console.log(`Found ${firstPage.total_count} marketplace.json files`);\n\n let indexed = 0;\n let failed = 0;\n let invalid = 0;\n let pageSearchFailures = 0;\n\n // Calculate total pages (max 10 pages = 1000 results)\n const maxPages = Math.min(Math.ceil(firstPage.total_count / 100), 10);\n if (firstPage.total_count > 1000) {\n console.warn(\n `Found ${firstPage.total_count} results. GitHub limits to 1000, only first 1000 will be indexed.`,\n );\n }\n\n const allPlugins: Array[number]> =\n [];\n\n // Process all pages\n for (let page = 1; page <= maxPages; page++) {\n console.log(`\\nProcessing page ${page}/${maxPages}...`);\n\n let searchResults: GitHubSearchResponse;\n try {\n searchResults = page === 1\n ? firstPage\n : await github.searchCode(searchQuery, page);\n } catch (error) {\n console.error(\n `Failed to fetch search results for page ${page}:`,\n error,\n );\n pageSearchFailures++;\n continue;\n }\n\n console.log(`Processing ${searchResults.items.length} items...`);\n\n const pageResults = await Promise.allSettled(\n searchResults.items.map(async (item) => {\n const [metadata, content] = await Promise.all([\n await github.fetchRepoMetadata(\n item.repository.full_name,\n ),\n await github.fetchFileContent(\n item.repository.full_name,\n item.path,\n ),\n ]);\n\n if (!content) {\n console.warn(`No content found for ${item.repository.full_name}`);\n throw new Error(\n `No content found for ${item.repository.full_name}`,\n );\n }\n\n const marketplace = parseMarketplace(\n content,\n item.repository.full_name,\n );\n if (!marketplace) {\n console.warn(\n `Invalid marketplace.json from ${item.repository.full_name}`,\n );\n return {\n status: \"invalid\" as const,\n repo: item.repository.full_name,\n };\n }\n\n const dbPlugins = transformToDbPlugins(\n marketplace,\n item.repository.owner.login,\n item.repository.html_url,\n metadata.stargazers_count,\n );\n\n return {\n status: \"success\" as const,\n repo: item.repository.full_name,\n plugins: dbPlugins,\n };\n }),\n );\n\n for (const result of pageResults) {\n if (result.status === \"rejected\") {\n console.error(`Failed to process item:`, result.reason);\n failed++;\n continue;\n }\n\n const value = result.value;\n if (value.status === \"invalid\") {\n invalid++;\n continue;\n }\n\n if (value.status === \"success\") {\n allPlugins.push(...value.plugins);\n }\n }\n }\n\n for (const chunk of chunkArray(allPlugins, 100)) {\n try {\n await db\n .insert(plugins)\n .values(chunk)\n .onConflictDoUpdate({\n target: [plugins.namespace, plugins.name],\n set: {\n description: sql`excluded.description`,\n keywords: sql`excluded.keywords`,\n category: sql`excluded.category`,\n version: sql`excluded.version`,\n author: sql`excluded.author`,\n metadata: sql`excluded.metadata`,\n gitUrl: sql`excluded.git_url`,\n stars: sql`excluded.stars`,\n skills: sql`COALESCE(excluded.skills, plugins.skills)`,\n updatedAt: sql`CURRENT_TIMESTAMP`,\n },\n })\n .run();\n\n indexed += chunk.length;\n } catch (dbError) {\n console.error(`Failed to upsert plugin batch:`, dbError);\n failed += chunk.length;\n }\n }\n\n const endTime = Date.now();\n const duration = endTime - startTime;\n\n console.log(`Plugins Indexer Completed:`, {\n duration: `${duration}ms`,\n indexed,\n failed,\n invalid,\n pageSearchFailures,\n });\n } catch (e) {\n console.error(\"Indexer CRON job failed:\", e);\n // so Val.town will notify\n throw e;\n }\n}",[],[223,257],"0cfc4b84-0a11-11f1-8914-42dde27851f2","parser.ts","019c5f00-3df9-72b9-8e7e-ecd8a1134b55",["D",1771120376884],"import { z } from \"npm:zod@4.1.12\";\n\n/**\n * Zod schema for marketplace.json structure\n */\nconst PluginSchema = z\n .object({\n name: z.string().min(1),\n description: z.string().optional(),\n version: z.string().optional(),\n category: z.string().optional(),\n skills: z.array(z.string()).optional(),\n })\n .passthrough(); // Allow ALL other fields\n\nconst MarketplaceSchema = z\n .object({\n name: z.string().min(1),\n plugins: z.array(PluginSchema),\n })\n .passthrough();\n\nexport type MarketplaceData = z.infer;\nexport type PluginData = z.infer;\n\n/**\n * Parse and validate marketplace.json content\n * Handles both standard marketplace format and non-standard single plugin format\n */\nexport function parseMarketplace(\n jsonString: string,\n repoFullName: string,\n): MarketplaceData | null {\n try {\n const jsonData = JSON.parse(jsonString);\n\n // First, try standard marketplace.json format\n const result = MarketplaceSchema.safeParse(jsonData);\n\n if (result.success) {\n return result.data;\n }\n\n // If standard format fails, check if this is a single plugin definition\n // (no marketplace wrapper, just a plugin object)\n if (jsonData.name && !jsonData.plugins) {\n console.log(\n `Detected single plugin format in ${repoFullName}, attempting to wrap...`,\n );\n\n // Try to wrap it as a marketplace with one plugin\n const wrapped = {\n name: jsonData.name,\n owner: jsonData.author || jsonData.owner || { name: \"Unknown\" },\n plugins: [jsonData],\n };\n\n const wrappedResult = MarketplaceSchema.safeParse(wrapped);\n if (wrappedResult.success) {\n console.log(`Successfully wrapped single plugin from ${repoFullName}`);\n return wrappedResult.data;\n }\n }\n\n // Still failed - log the errors\n console.error(`Invalid marketplace.json from ${repoFullName}:`);\n console.error(\n `Validation errors: ${JSON.stringify(result.error.errors, null, 2)}`,\n );\n return null;\n } catch (error) {\n console.error(\n `Failed to parse JSON from ${repoFullName}:`,\n error instanceof Error ? error.message : String(error),\n );\n console.error(`Stack: ${error instanceof Error ? error.stack : \"\"}`);\n return null;\n }\n}\n\n/**\n * Transform marketplace data into database plugin records\n */\nexport function transformToDbPlugins(\n marketplace: MarketplaceData,\n repoOwner: string,\n repoUrl: string,\n repoStars: number = 0,\n): Array<{\n id: string;\n name: string;\n namespace: string;\n gitUrl: string;\n description: string;\n keywords: string;\n skills: string;\n category: string | null;\n version: string;\n author: string;\n metadata: string;\n stars: number;\n verified: boolean;\n downloads: number;\n createdAt: string;\n updatedAt: string;\n}> {\n return marketplace.plugins.map((plugin: any) => {\n // Safe getter helper\n const get = (obj: any, key: string, defaultValue: any = null) => {\n return obj?.[key] ?? defaultValue;\n };\n\n // Simple author extraction - just get a string\n let author = \"Unknown\";\n const pluginAuthor = get(plugin, \"author\");\n if (typeof pluginAuthor === \"string\") {\n author = pluginAuthor;\n } else if (pluginAuthor?.name) {\n author = pluginAuthor.name;\n } else {\n const marketplaceOwner = get(marketplace, \"owner\");\n const marketplacePublisher = get(marketplace, \"publisher\");\n if (typeof marketplacePublisher === \"string\") {\n author = marketplacePublisher;\n } else if (marketplaceOwner?.name) {\n author = marketplaceOwner.name;\n } else {\n author = repoOwner;\n }\n }\n\n // Simple version - use plugin version, fallback to marketplace, then default\n let version = get(plugin, \"version\");\n if (!version || !/^\\d+\\.\\d+\\.\\d+/.test(version)) {\n const marketplaceVersion = get(marketplace, \"version\");\n version = marketplaceVersion && /^\\d+\\.\\d+\\.\\d+/.test(marketplaceVersion)\n ? marketplaceVersion\n : \"1.0.0\";\n }\n\n // Extract keywords from various sources\n const keywords = get(plugin, \"keywords\", []);\n const tags = get(plugin, \"tags\", []);\n const allKeywords = [...keywords, ...tags].filter(\n (k: any) => typeof k === \"string\",\n );\n\n // Everything else goes into metadata\n const metadata = {\n homepage: get(plugin, \"homepage\"),\n repository: get(plugin, \"repository\"),\n license: get(plugin, \"license\"),\n commands: get(plugin, \"commands\"),\n agents: get(plugin, \"agents\"),\n mcpServers: get(plugin, \"mcpServers\"),\n // Include any other fields that might be useful later\n ...(plugin.bugs && { bugs: plugin.bugs }),\n ...(plugin.features && { features: plugin.features }),\n ...(plugin.requirements && { requirements: plugin.requirements }),\n };\n\n return {\n id: crypto.randomUUID(),\n name: plugin.name,\n namespace: `@${repoOwner}/${marketplace.name}`,\n gitUrl: repoUrl,\n description: get(plugin, \"description\", \"\"),\n keywords: JSON.stringify(allKeywords),\n category: get(plugin, \"category\"),\n skills: JSON.stringify(plugin.skills || []),\n version: version,\n author: author,\n metadata: JSON.stringify(metadata),\n stars: repoStars,\n verified: false,\n downloads: 0,\n createdAt: new Date().toISOString(),\n updatedAt: new Date().toISOString(),\n };\n });\n}",[],[223,264],"0cfc4328-0a11-11f1-8914-42dde27851f2","indexer-query-tester.ts","019c5f00-3df9-72b9-8e7e-eb8a23962c03",["D",1771120376884],"import { GitHubClient } from \"../clients/github.ts\";\n\nconst github = new GitHubClient();\n\n// const testQueries = [\n// 'filename:SKILL.md path:skills/ language:Markdown \"---\" \"name:\" \"description:\"',\n// 'filename:SKILL.md language:Markdown \"---\" \"name:\" \"description:\" pushed:>=2025-10-14',\n// 'filename:SKILL.md language:Markdown size:100..5000 \"---\" \"name:\" \"description:\"',\n// 'filename:SKILL.md language:Markdown size:5001..20000 \"---\" \"name:\" \"description:\"',\n// ]\nconst searchQuery = \"repo:vercel-labs/agent-skills filename:SKILL.md\";\nconsole.log(`Searching: ${searchQuery}`);\n\nconst firstPage = await github.searchCode(searchQuery);\n// console.log(`Found ${firstPage.total_count} SKILL.md files`);\n\n// Start with a broad plausible range for SKILL.md\n// const MIN_SIZE = 500;\n// const MAX_SIZE = 150_000;\n// const BASE_QUERY = `filename:SKILL.md language:Markdown \"---\" \"name:\" \"description:\"`;\n\n// const buckets = await github.findBuckets(BASE_QUERY, {min: MIN_SIZE, max: MAX_SIZE, maxPerBucket: 950, minSpan: 200});\n// console.log(`Found ${buckets.length} buckets`, buckets.map(console.log));\n\n// Fullname: Kamalnrf/claude-plugins, path: patches/giget@2.0.0.patch\n// const fileContent = await fetch(\n// \"https://raw.githubusercontent.com/Kamalnrf/claude-plugins/refs/heads/main/CLAUDE.md\",\n// ).then((res) => res.text());\n\n// console.log(\"First Commit At =>\", fileContent);",[],[223,271],"0cfc41c0-0a11-11f1-8914-42dde27851f2","clients","019c5f00-3df9-72b9-8e7e-e66701c0104d",["D",1771120376884],[283,284,285],[278],{"_36":300,"_13":301,"_62":302,"_64":279,"_66":67,"_68":34,"_69":70,"_38":303,"_17":32,"_72":97,"_74":304,"_75":51,"_76":-5,"_99":305,"_18":306},{"_36":293,"_13":294,"_62":295,"_64":279,"_66":67,"_68":34,"_69":70,"_38":296,"_17":32,"_72":97,"_74":297,"_75":51,"_76":-5,"_99":298,"_18":299},{"_36":286,"_13":287,"_62":288,"_64":279,"_66":67,"_68":34,"_69":70,"_38":289,"_17":32,"_72":97,"_74":290,"_75":51,"_76":-5,"_99":291,"_18":292},"0cfc3d74-0a11-11f1-8914-42dde27851f2","sqlite.ts","019c5f00-3df9-72b9-8e7e-e2245011e9fd",["D",1771120376884],"import { createClient } from \"https://esm.sh/@libsql/client@0.6.0/web\";\nimport { sqlite as vtSQLite } from \"https://esm.town/v/std/sqlite2/main.tsx\";\n\n// Optional if you want to use a different sqlite db\nconst URL = Deno.env.get(\"DB_API_URL\");\nconst TOKEN = Deno.env.get(\"DB_API_TOKEN\");\n\nexport const sqlite = (() => {\n if (URL && TOKEN) {\n return createClient({\n url: URL,\n authToken: TOKEN,\n });\n }\n\n return vtSQLite;\n})();",[],[278,287],"0cfc0b4c-0a11-11f1-8914-42dde27851f2","github.ts","019c5f00-3df9-72b9-8e7e-dce4d5135ae3",["D",1771120376884],"import {\n getToken,\n syncTokenHealth,\n} from \"https://esm.town/v/kamalnrf/token-pool/lib/pool.ts\";\nimport { z } from \"npm:zod@3.24.2\";\n\nconst REQUESTS_PER_INTERVAL = 80;\nconst INTERVAL_MS = 1000;\n\nexport type Bucket = {\n min: number;\n max: number;\n count: number;\n};\n\n/**\n * GitHub code search response types\n */\nconst GitHubSearchItemSchema = z.object({\n name: z.string(),\n path: z.string(),\n sha: z.string(),\n url: z.string().url(),\n html_url: z.string().url(),\n repository: z.object({\n id: z.number(),\n full_name: z.string(),\n html_url: z.string().url(),\n owner: z.object({\n login: z.string(),\n }),\n }),\n});\n\nconst GitHubRepoSchema = z.object({\n id: z.number(),\n full_name: z.string(),\n stargazers_count: z.number(),\n updated_at: z.string(),\n html_url: z.string().url(),\n fork: z.boolean(),\n default_branch: z.string(),\n});\n\nconst GitHubTreeSchema = z.object({\n sha: z.string(),\n truncated: z.boolean().optional(),\n tree: z.array(\n z.object({\n path: z.string(),\n type: z.enum([\"blob\", \"tree\", \"commit\"]),\n sha: z.string().optional(),\n size: z.number().optional(),\n }),\n ),\n});\n\nexport type GitHubTree = z.infer;\n\nexport type ListRepoFilesResult = {\n repoFullName: string;\n repoHtmlUrl: string;\n defaultBranch: string;\n files: Array<{ path: string; type: \"blob\" | \"tree\" }>;\n truncated: boolean;\n};\n\nconst GitHubSearchResponseSchema = z.object({\n total_count: z.number(),\n incomplete_results: z.boolean(),\n items: z.array(GitHubSearchItemSchema),\n});\n\nexport type GitHubSearchItem = z.infer;\nexport type GitHubSearchResponse = z.infer;\nexport type GitHubRepo = z.infer;\n\n/**\n * GitHub API client with rate limit handling\n */\nexport class GitHubClient {\n private baseUrl = \"https://api.github.com\";\n private apiVersion = \"2022-11-28\";\n private tokens = REQUESTS_PER_INTERVAL;\n private lastRefill = Date.now();\n\n /**\n * Search for files matching a path pattern\n */\n async searchCode(\n query: string,\n page: number = 1,\n ): Promise {\n const url = `${this.baseUrl}/search/code?q=${\n encodeURIComponent(query)\n }&per_page=100&page=${page}`;\n\n const response = await this.fetchWithRetry(url);\n\n // Check rate limit\n const remaining = parseInt(\n response.headers.get(\"x-ratelimit-remaining\") || \"10\",\n );\n const resetTime = parseInt(\n response.headers.get(\"x-ratelimit-reset\") || \"0\",\n );\n\n console.log(\n `GitHub rate limit: ${remaining}/10 remaining, resets at ${\n new Date(resetTime * 1000).toISOString()\n }`,\n );\n\n if (remaining <= 2) {\n const waitSeconds = resetTime - Math.floor(Date.now() / 1000) + 5;\n if (waitSeconds > 0) {\n console.warn(\n `Rate limit low. Waiting ${waitSeconds}s before continuing...`,\n );\n await this.sleep(waitSeconds * 1000);\n }\n }\n\n if (!response.ok) {\n throw new Error(\n `GitHub search failed: ${response.status} ${response.statusText}`,\n );\n }\n\n const data = await response.json();\n return GitHubSearchResponseSchema.parse(data);\n }\n\n /**\n * Fetch raw file content from repository\n */\n async fetchFileContent(repoFullName: string, path: string): Promise {\n const url = `${this.baseUrl}/repos/${repoFullName}/contents/${path}`;\n\n const response = await this.fetchWithRetry(url);\n\n if (!response.ok) {\n throw new Error(\n `Failed to fetch ${path} from ${repoFullName}: ${response.status}`,\n );\n }\n\n const data = await response.json();\n\n // Decode base64 content\n if (data.encoding === \"base64\" && data.content) {\n return atob(data.content.replace(/\\n/g, \"\"));\n }\n\n throw new Error(`Unexpected content encoding: ${data.encoding}`);\n }\n\n /**\n * Fetch repository metadata (stars, updated_at, etc.)\n */\n async fetchRepoMetadata(repoFullName: string): Promise {\n const url = `${this.baseUrl}/repos/${repoFullName}`;\n\n const response = await this.fetchWithRetry(url);\n\n if (!response.ok) {\n throw new Error(\n `Failed to fetch repo metadata for ${repoFullName}: ${response.status}`,\n );\n }\n\n const data = await response.json();\n return GitHubRepoSchema.parse(data);\n }\n\n async findBuckets(\n baseQuery: string,\n {\n min,\n max,\n maxPerBucket = 950,\n minSpan = 100,\n }: {\n min: number;\n max: number;\n maxPerBucket?: number;\n minSpan?: number;\n },\n ): Promise> {\n const query = `${baseQuery} size:${min}..${max}`;\n const count = (await this.searchCode(query)).total_count ?? 0;\n console.log(\"Query =>\", {\n query,\n count,\n });\n\n if (count === 0) return [];\n\n if (count <= maxPerBucket || (max - min) <= minSpan) {\n return [{ min, max, count }];\n }\n\n // Split the range and recurse\n const mid = Math.floor((min + max) / 2);\n const left = await this.findBuckets(\n baseQuery,\n { min: min, max: mid, maxPerBucket, minSpan },\n );\n const right = await this.findBuckets(\n baseQuery,\n { min: mid + 1, max: max, maxPerBucket, minSpan },\n );\n return [...left, ...right];\n }\n\n /**\n * Fetch with exponential backoff retry\n */\n private async fetchWithRetry(\n url: string,\n maxRetries: number = 3,\n ): Promise {\n // Throttling requests to Github before any attempts\n await this.waitForToken();\n\n let lastError: Error | null = null;\n\n for (let attempt = 0; attempt <= maxRetries; attempt++) {\n try {\n const t = await getToken();\n if (!t?.token) {\n throw new Error(\"No Github token found\");\n }\n\n const response = await fetch(url, {\n headers: {\n Accept: \"application/vnd.github+json\",\n Authorization: `Bearer ${t.token}`,\n \"X-GitHub-Api-Version\": this.apiVersion,\n },\n });\n\n await syncTokenHealth(t.id, response);\n // Handle rate limiting with Retry-After\n if (response.status === 429 || response.status === 403) {\n console.log(\n `Rate limited. with ${t.id}, retrying with new token`,\n );\n continue;\n }\n\n return response;\n } catch (error) {\n lastError = error as Error;\n\n if (attempt < maxRetries) {\n const delay = Math.min(1000 * Math.pow(2, attempt), 10000);\n console.log(\n `Request failed, retrying in ${delay}ms... (attempt ${\n attempt + 1\n }/${maxRetries})`,\n );\n await this.sleep(delay);\n }\n }\n }\n\n throw lastError || new Error(\"Request failed after retries\");\n }\n\n private async waitForToken(): Promise {\n const now = Date.now();\n const elapsed = now - this.lastRefill;\n if (elapsed >= INTERVAL_MS) {\n const intervals = Math.floor(elapsed / INTERVAL_MS);\n this.tokens = Math.min(\n REQUESTS_PER_INTERVAL,\n this.tokens + intervals * REQUESTS_PER_INTERVAL,\n );\n this.lastRefill = now;\n }\n\n if (this.tokens > 0) {\n this.tokens--;\n return;\n }\n\n const delay = Math.max(0, INTERVAL_MS - (Date.now() - this.lastRefill));\n await this.sleep(delay);\n // After waiting, consume a token\n if (this.tokens > 0) this.tokens--;\n else this.tokens = Math.max(0, this.tokens - 1);\n }\n\n private sleep(ms: number): Promise {\n return new Promise((resolve) => setTimeout(resolve, ms));\n }\n\n /**\n * List all files in a repository using the Git Trees API\n */\n async listRepoFiles(\n repoFullName: string,\n ): Promise {\n const repo = await this.fetchRepoMetadata(repoFullName);\n const url =\n `${this.baseUrl}/repos/${repoFullName}/git/trees/${repo.default_branch}?recursive=1`;\n const response = await this.fetchWithRetry(url);\n\n if (!response.ok) {\n throw new Error(\n `Failed to fetch tree for ${repoFullName}: ${response.status}`,\n );\n }\n\n const data = await response.json();\n const tree = GitHubTreeSchema.parse(data);\n\n return {\n repoFullName: repo.full_name,\n repoHtmlUrl: repo.html_url,\n defaultBranch: repo.default_branch,\n files: tree.tree\n .filter((item) => item.type === \"blob\" || item.type === \"tree\")\n .map((item) => ({\n path: item.path,\n type: item.type as \"blob\" | \"tree\",\n })),\n truncated: tree.truncated ?? false,\n };\n }\n\n /**\n * Fetch the date when a file was first committed using GitHub REST API.\n * Uses Link header pagination to find the last page (oldest commit).\n * Returns ISO date string or null if unable to determine.\n */\n async fetchFirstCommitDate(\n repoFullName: string,\n filePath: string,\n ): Promise {\n try {\n // First request: get commits for this file with per_page=1 to check Link header\n const url = `${this.baseUrl}/repos/${repoFullName}/commits?path=${\n encodeURIComponent(filePath)\n }&per_page=1`;\n const response = await this.fetchWithRetry(url);\n\n if (!response.ok) {\n console.error(\n `Failed to fetch commits for ${repoFullName}/${filePath}: ${response.status}`,\n );\n return null;\n }\n\n // Parse Link header to find last page\n const linkHeader = response.headers.get(\"Link\");\n let lastPage = 1;\n\n if (linkHeader) {\n // Link header format: ; rel=\"next\", ; rel=\"last\"\n const lastMatch = linkHeader.match(\n /<[^>]+[?&]page=(\\d+)[^>]*>;\\s*rel=\"last\"/,\n );\n if (lastMatch) {\n lastPage = parseInt(lastMatch[1], 10);\n }\n }\n\n // If only one page, the first response already has the oldest commit\n if (lastPage === 1) {\n const commits = await response.json();\n if (commits.length === 0) {\n return null;\n }\n // Single page means this is also the oldest\n const oldestCommit = commits[0];\n return oldestCommit?.commit?.author?.date ||\n oldestCommit?.commit?.committer?.date || null;\n }\n\n // Fetch the last page to get the oldest commit\n const lastPageUrl = `${this.baseUrl}/repos/${repoFullName}/commits?path=${\n encodeURIComponent(filePath)\n }&per_page=1&page=${lastPage}`;\n const lastPageResponse = await this.fetchWithRetry(lastPageUrl);\n\n if (!lastPageResponse.ok) {\n console.error(\n `Failed to fetch last page of commits for ${repoFullName}/${filePath}: ${lastPageResponse.status}`,\n );\n return null;\n }\n\n const lastPageCommits = await lastPageResponse.json();\n if (lastPageCommits.length === 0) {\n return null;\n }\n\n const oldestCommit = lastPageCommits[0];\n return oldestCommit?.commit?.author?.date ||\n oldestCommit?.commit?.committer?.date || null;\n } catch (error) {\n console.error(\n `Failed to fetch first commit date for ${repoFullName}/${filePath}:`,\n error,\n );\n return null;\n }\n }\n}",[],[278,294],"0cfbfcd8-0a11-11f1-8914-42dde27851f2","cache.ts","019c5f00-3df9-72b9-8e7e-d935b55bd024",["D",1771120376884],"/**\n * Simple blob-based cache\n */\nimport { blob } from \"https://esm.town/v/std/blob\";\nimport { sqlite } from \"./sqlite.ts\";\n\ntype CachedValue = {\n value: T;\n updatedAt: number;\n};\n\n// 10 Mins\nconst DEFAULT_TTL_MS = 10 * 60 * 1000;\n\n/**\n * Get a cached value, or compute and cache it if expired/missing\n */\nexport async function getCached(\n key: string,\n compute: () => Promise,\n ttlMs: number = DEFAULT_TTL_MS,\n): Promise {\n const now = Date.now();\n\n try {\n const cached = await blob.getJSON>(key);\n if (cached && now - cached.updatedAt < ttlMs) {\n return cached.value;\n }\n } catch {\n // Cache miss or parse error\n }\n\n const value = await compute();\n await blob.setJSON(key, { value, updatedAt: now } as CachedValue);\n return value;\n}\n\n/**\n * Invalidate a cached value\n */\nexport async function invalidateCache(key: string): Promise {\n try {\n await blob.delete(key);\n } catch {\n // Ignore if not found\n }\n}\n\n// Cache keys\nexport const CACHE_KEYS = {\n ACTIVE_SKILLS_COUNT: \"cache:skills:count:active\",\n ACTIVE_PLUGINS_COUNT: \"cache:plugins:count:all\",\n SITEMAP_SKILLS: \"cache:sitemap:skills\",\n SITEMAP_PLUGINS: \"cache:sitemap:plugins\",\n} as const;\n\n// 1 day TTL for sitemap data\nconst SITEMAP_CACHE_TTL_MS = 24 * 60 * 60 * 1000;\n\nexport function getSitemapSkillsCacheKey(\n limit: number,\n offset: number,\n): string {\n return `${CACHE_KEYS.SITEMAP_SKILLS}:${offset}:${limit}`;\n}\n\nexport function getSitemapPluginsCacheKey(\n limit: number,\n offset: number,\n): string {\n return `${CACHE_KEYS.SITEMAP_PLUGINS}:${offset}:${limit}`;\n}\n\nexport { SITEMAP_CACHE_TTL_MS };\n\n/**\n * Get cached count of active skills\n * Uses FTS table which only contains active skills\n */\nexport async function getActiveSkillsCount(): Promise {\n // This is since a sizable about new skills get indexed only once every hour\n const ONE_HOUR_MS = 60 * 60 * 1000;\n\n return getCached(\n CACHE_KEYS.ACTIVE_SKILLS_COUNT,\n async () => {\n const result = await sqlite.execute({\n sql: `SELECT COUNT(*) as count FROM skills_fts`,\n args: [],\n });\n return Number(result.rows[0]?.[0] ?? 0);\n },\n ONE_HOUR_MS,\n );\n}\n\n/**\n * Get cached count of all plugins\n */\nexport async function getPluginsCount(): Promise {\n // This is since new plugins get indexed only once every hour\n const ONE_HOUR_MS = 60 * 60 * 1000;\n\n return getCached(\n CACHE_KEYS.ACTIVE_PLUGINS_COUNT,\n async () => {\n const result = await sqlite.execute({\n sql: `SELECT COUNT(*) as count FROM plugins`,\n args: [],\n });\n return Number(result.rows[0]?.[0] ?? 0);\n },\n ONE_HOUR_MS,\n );\n}",[],[278,301],"0cfbfb48-0a11-11f1-8914-42dde27851f2","api","019c5f00-3df9-72b9-8e7e-d7101610bec4",["D",1771120376884],[313,314,315,316,317,318,319],[308],{"_36":362,"_13":363,"_62":364,"_64":309,"_66":67,"_68":34,"_69":70,"_38":365,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":366,"_18":367},{"_36":355,"_13":356,"_62":357,"_64":309,"_66":67,"_68":34,"_69":70,"_38":358,"_17":32,"_72":97,"_74":359,"_75":51,"_76":-5,"_99":360,"_18":361},{"_36":348,"_13":349,"_62":350,"_64":309,"_66":67,"_68":34,"_69":70,"_38":351,"_17":32,"_72":97,"_74":352,"_75":51,"_76":-5,"_99":353,"_18":354},{"_36":341,"_13":342,"_62":343,"_64":309,"_66":67,"_68":34,"_69":70,"_38":344,"_17":32,"_72":97,"_74":345,"_75":51,"_76":-5,"_99":346,"_18":347},{"_36":334,"_13":335,"_62":336,"_64":309,"_66":67,"_68":34,"_69":70,"_38":337,"_17":32,"_72":97,"_74":338,"_75":51,"_76":-5,"_99":339,"_18":340},{"_36":327,"_13":328,"_62":329,"_64":309,"_66":67,"_68":34,"_69":70,"_38":330,"_17":32,"_72":97,"_74":331,"_75":51,"_76":-5,"_99":332,"_18":333},{"_36":320,"_13":321,"_62":322,"_64":309,"_66":67,"_68":34,"_69":70,"_38":323,"_17":32,"_72":97,"_74":324,"_75":51,"_76":-5,"_99":325,"_18":326},"0cfbc61e-0a11-11f1-8914-42dde27851f2","skills.ts","019c5f00-3df9-72b9-8e7e-d064e2958b11",["D",1771120376884],"/**\n * Skills API endpoints\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, asc, desc, eq, sql } from \"npm:drizzle-orm@0.44.2\";\nimport { sqlite } from \"../clients/sqlite.ts\";\nimport { db } from \"../database/db.ts\";\nimport { skills, skillStats } from \"../database/schema.ts\";\nimport { getLogger } from \"https://esm.town/v/kamalnrf/axiom-logger/hono.ts\";\nimport { getActiveSkillsCount } from \"../clients/cache.ts\";\nimport { sanitizeFtsQuery } from \"../utils.ts\";\nimport { GitHubClient } from \"../clients/github.ts\";\nimport { discoverSkillsInRepo } from \"../lib/skills/discover.ts\";\nimport { indexSkillFiles } from \"../lib/skills/index-skill-files.ts\";\n\nconst github = new GitHubClient();\n\nconst ALLOWED_SORT_COLUMNS = new Set([\n \"downloads\",\n \"stars\",\n]);\n\nfunction getSkillColumn(field: string) {\n switch (field) {\n case \"downloads\":\n return skills.installs;\n case \"stars\":\n return skills.stars;\n default:\n return null;\n }\n}\n\n/**\n * GET /api/skills/:owner/:repo/:skillName\n * Get skill info (read-only, no install increment)\n */\nexport async function getSkill(c: Context) {\n const { owner, repo, skillName } = c.req.param();\n\n if (!owner || !repo || !skillName) {\n return c.json({ error: \"Missing owner, repo, or skill name\" }, 400);\n }\n\n const namespace = `${owner}/${repo}/${skillName}`;\n\n const skill = await db\n .select()\n .from(skills)\n .where(eq(skills.namespace, namespace))\n .get();\n\n if (!skill) {\n return c.json({\n error: `Skill ${namespace} not found`,\n }, 404);\n }\n\n const metadata = skill.metadata ? JSON.parse(skill.metadata) : {};\n\n return c.json({\n id: skill.id,\n name: skill.name,\n namespace: skill.namespace,\n sourceUrl: skill.sourceUrl,\n sourceType: skill.sourceType,\n description: skill.description,\n version: skill.version,\n dependencies: skill.dependencies,\n author: skill.author,\n stars: skill.stars,\n installs: skill.installs,\n verified: Boolean(skill.verified),\n metadata,\n createdAt: skill.createdAt,\n updatedAt: skill.updatedAt,\n });\n}\n\n/**\n * POST /api/skills/:owner/:repo/:skillName/install\n * Track skill installation - auto-indexes unknown skills\n */\nexport async function installSkill(c: Context) {\n const logger = getLogger(c);\n const { owner, repo, skillName } = c.req.param();\n logger.info(\"resolved values\", { owner, repo, skillName });\n\n if (!owner || !repo || !skillName) {\n return c.json({ error: \"Missing owner, repo, or skill name\" }, 400);\n }\n \n const namespace = `${owner.startsWith('@') ? owner : `@${owner}`}/${repo}/${skillName}`;\n\n let skill = await db\n .select()\n .from(skills)\n .where(eq(skills.namespace, namespace))\n .get();\n\n // If not found, try to discover and index from GitHub\n if (!skill) {\n logger.info(\"Skill not found, attempting to index from GitHub\", { namespace });\n \n try {\n const repoFullName = `${owner}/${repo}`;\n const discovered = await discoverSkillsInRepo(github, repoFullName);\n \n // Find the specific skill in discovered results\n const targetSkillRef = discovered.skills.find((s) => {\n const skillDir = s.path.replace(/\\/SKILL\\.md$/, \"\");\n const name = skillDir.split(\"/\").pop() ?? \"\";\n return name === skillName;\n });\n\n if (targetSkillRef) {\n // Index just this skill\n const indexResult = await indexSkillFiles([targetSkillRef], { github, log: logger, dryRun: false });\n logger.info(\"Index result\", { indexed: indexResult.indexed, failed: indexResult.failed });\n \n // Re-query after indexing\n skill = await db\n .select()\n .from(skills)\n .where(eq(skills.namespace, namespace))\n .get();\n }\n } catch (error) {\n const message = error instanceof Error ? error.message : String(error);\n logger.warn(\"Failed to auto-index skill\", { error: message, namespace });\n // Continue to 404 if indexing fails\n }\n }\n\n if (!skill) {\n return c.json({\n error: `Skill ${namespace} not found`,\n }, 404);\n }\n\n try {\n await incrementInstalls(skill.id);\n return c.json({ success: true });\n } catch (error) {\n logger.error(\"Failed to increment installs:\", { error });\n return c.json({ error: \"Failed to track installation\" }, 500);\n }\n}\n\n/**\n * GET /api/skills/search?q=query&limit=20&offset=0\n * Search skills - uses FTS5 for text queries, Drizzle for empty queries\n */\nexport async function searchSkills(c: Context) {\n const query = c.req.query(\"q\") || \"\";\n // Cap limit at 200 to prevent excessive row reads\n const limit = Math.min(parseInt(c.req.query(\"limit\") || \"20\"), 200);\n const offset = parseInt(c.req.query(\"offset\") || \"0\");\n const orderByParam = c.req.query(\"orderBy\");\n const orderParam = c.req.query(\"order\") ?? \"desc\";\n\n if (orderByParam && !ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n return c.json({ error: \"Invalid orderBy parameter\" }, 400);\n }\n\n // Empty query = list active skills (no FTS needed, use Drizzle)\n if (!query.trim()) {\n return listActiveSkills(c, { limit, offset, orderByParam, orderParam });\n }\n\n // Text search = use FTS5 with raw SQL (Drizzle doesn't support FTS well)\n return searchSkillsWithFTS(c, {\n query,\n limit,\n offset,\n orderByParam,\n orderParam,\n });\n}\n\n/**\n * List active skills without search query\n * Uses Drizzle ORM - optimized with partial index on (deleted=0, duplicate=0)\n */\nasync function listActiveSkills(\n c: Context,\n { limit, offset, orderByParam, orderParam }: {\n limit: number;\n offset: number;\n orderByParam: string | undefined;\n orderParam: string;\n },\n) {\n const baseCondition = and(\n eq(skills.deleted, false),\n eq(skills.duplicate, false),\n );\n\n // Partial select - only columns needed for API response\n const selectedColumns = {\n id: skills.id,\n name: skills.name,\n namespace: skills.namespace,\n sourceUrl: skills.sourceUrl,\n description: skills.description,\n version: skills.version,\n dependencies: skills.dependencies,\n author: skills.author,\n stars: skills.stars,\n installs: skills.installs,\n metadata: skills.metadata,\n createdAt: skills.createdAt,\n updatedAt: skills.updatedAt,\n };\n\n // Determine sort order\n let orderByClause;\n if (orderByParam && ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n const column = getSkillColumn(orderByParam);\n const sortDirection = orderParam === \"asc\" ? asc : desc;\n orderByClause = column\n ? [sortDirection(column)]\n : [desc(skills.installs), desc(skills.stars)];\n } else {\n orderByClause = [desc(skills.installs), desc(skills.stars)];\n }\n\n const total = await getActiveSkillsCount();\n\n const results = await db\n .select(selectedColumns)\n .from(skills)\n .where(baseCondition)\n .orderBy(...orderByClause)\n .limit(limit)\n .offset(offset)\n .all();\n\n const formattedSkills = results.map((skill) => ({\n id: skill.id,\n name: skill.name,\n namespace: skill.namespace,\n sourceUrl: skill.sourceUrl,\n description: skill.description,\n version: skill.version,\n dependencies: skill.dependencies,\n author: skill.author,\n stars: skill.stars,\n installs: skill.installs,\n metadata: skill.metadata ? JSON.parse(skill.metadata) : {},\n createdAt: skill.createdAt,\n updatedAt: skill.updatedAt,\n }));\n\n return c.json({\n skills: formattedSkills,\n total,\n limit,\n offset,\n });\n}\n\n/**\n * FTS5 full-text search with BM25 ranking\n * Uses raw SQL because Drizzle doesn't support FTS5 MATCH syntax\n */\nasync function searchSkillsWithFTS(\n c: Context,\n {\n query,\n limit,\n offset,\n orderByParam,\n orderParam,\n }: {\n query: string;\n limit: number;\n offset: number;\n orderByParam: string | undefined;\n orderParam: string;\n },\n) {\n // Sanitize and prepare FTS5 query\n const ftsQuery = sanitizeFtsQuery(query);\n\n if (!ftsQuery) {\n // If sanitization removed everything, fall back to listing\n return listActiveSkills(c, { limit, offset, orderByParam, orderParam });\n }\n\n try {\n // Count total matches - FTS table only contains active skills\n const countResult = await sqlite.execute({\n sql: `SELECT COUNT(*) as count FROM skills_fts WHERE skills_fts MATCH ?`,\n args: [ftsQuery],\n });\n const total = Number(countResult.rows[0]?.[0] ?? 0);\n\n // Build ORDER BY clause\n let orderClause: string;\n if (orderByParam && ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n const direction = orderParam === \"asc\" ? \"ASC\" : \"DESC\";\n const column = orderByParam === \"downloads\"\n ? \"skills.installs\"\n : \"skills.stars\";\n orderClause = `${column} ${direction}`;\n } else {\n // Combined score: BM25 relevance + popularity boost\n orderClause =\n `(bm25(skills_fts, 0, 10.0, 5.0, 1.0, 1.0) - (ln(skills.installs + 1) * 2.0) - (ln(skills.stars + 1) * 1.0)) ASC`;\n }\n\n // Execute FTS5 search with raw SQL\n const results = await sqlite.execute({\n sql: `\n SELECT\n skills.id,\n skills.name,\n skills.namespace,\n skills.source_url,\n skills.description,\n skills.version,\n skills.dependencies,\n skills.author,\n skills.stars,\n skills.installs,\n skills.metadata,\n skills.created_at,\n skills.updated_at\n FROM skills_fts\n JOIN skills ON skills.rowid = skills_fts.rowid\n WHERE skills_fts MATCH ?\n ORDER BY ${orderClause}\n LIMIT ? OFFSET ?\n `,\n args: [ftsQuery, limit, offset],\n });\n\n const formattedSkills = results.rows.map((row: any) => ({\n id: row[0],\n name: row[1],\n namespace: row[2],\n sourceUrl: row[3],\n description: row[4],\n version: row[5],\n dependencies: row[6],\n author: row[7],\n stars: row[8],\n installs: row[9],\n metadata: row[10] ? JSON.parse(row[10]) : {},\n createdAt: row[11],\n updatedAt: row[12],\n }));\n\n return c.json({\n skills: formattedSkills,\n total,\n limit,\n offset,\n });\n } catch (error) {\n console.error(\"FTS5 search failed:\", error);\n return c.json({\n error: \"Search temporarily unavailable. Please try again.\",\n details: String(error),\n }, 503);\n }\n}\n\n/**\n * GET /api/skills/:owner/:repo/:skillName/stats\n * Get skill statistics\n */\nexport async function getSkillStats(c: Context) {\n const { owner, repo, skillName } = c.req.param();\n const namespace = `@${owner}/${repo}/${skillName}`;\n\n const result = await db\n .select({\n skill: skills,\n stats: skillStats,\n })\n .from(skills)\n .leftJoin(skillStats, eq(skills.id, skillStats.skillId))\n .where(eq(skills.namespace, namespace))\n .get();\n\n if (!result || !result.skill) {\n return c.json({\n error: `Skill ${namespace} not found`,\n }, 404);\n }\n\n return c.json({\n installs: {\n total: result.stats?.installsTotal || 0,\n week: result.stats?.installsWeek || 0,\n month: result.stats?.installsMonth || 0,\n },\n lastInstalled: result.stats?.lastInstalledAt || null,\n });\n}\n\n/**\n * Increment install statistics for a skill\n * Uses upsert to avoid extra SELECT query\n */\nasync function incrementInstalls(skillId: string) {\n const now = new Date().toISOString();\n\n await db\n .update(skills)\n .set({ installs: sql`${skills.installs} + 1` })\n .where(eq(skills.id, skillId));\n\n await db\n .insert(skillStats)\n .values({\n skillId,\n installsTotal: 1,\n installsWeek: 1,\n installsMonth: 1,\n lastInstalledAt: now,\n })\n .onConflictDoUpdate({\n target: skillStats.skillId,\n set: {\n installsTotal: sql`${skillStats.installsTotal} + 1`,\n installsWeek: sql`${skillStats.installsWeek} + 1`,\n installsMonth: sql`${skillStats.installsMonth} + 1`,\n lastInstalledAt: now,\n },\n });\n}",[],[308,321],"0cfbb444-0a11-11f1-8914-42dde27851f2","sitemap.ts","019c5f00-3df8-778f-befc-c7b23059895f",["D",1771120376884],"/**\n * Sitemap endpoint for SEO\n * Returns minimal data needed for sitemap generation with pagination\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, desc, eq } from \"npm:drizzle-orm@0.44.2\";\nimport { db } from \"../database/db.ts\";\nimport { plugins, skills } from \"../database/schema.ts\";\nimport {\n getActiveSkillsCount,\n getCached,\n getPluginsCount,\n getSitemapPluginsCacheKey,\n getSitemapSkillsCacheKey,\n SITEMAP_CACHE_TTL_MS,\n} from \"../clients/cache.ts\";\n\nconst BASE_URL = \"https://claude-plugins.dev\";\nconst MAX_SITEMAP_LIMIT = 50000;\nconst DEFAULT_LIMIT = 50000;\n\ntype SitemapUrl = {\n loc: string;\n lastmod: string | undefined;\n};\n\ntype SitemapResponse = {\n urls: SitemapUrl[];\n total: number;\n limit: number;\n offset: number;\n};\n\n/**\n * GET /api/sitemap/skills\n * Returns paginated skills for sitemap generation (cached for 1 day)\n * Query params:\n * - limit: max 50000 (default 50000)\n * - offset: default 0\n */\nexport async function getSkillsSitemap(c: Context) {\n const limit = Math.min(\n parseInt(c.req.query(\"limit\") ?? String(DEFAULT_LIMIT)),\n MAX_SITEMAP_LIMIT,\n );\n const offset = parseInt(c.req.query(\"offset\") || \"0\");\n\n const result = await getCached(\n getSitemapSkillsCacheKey(limit, offset),\n async () => {\n const total = await getActiveSkillsCount();\n\n const results = await db\n .select({\n namespace: skills.namespace,\n updatedAt: skills.updatedAt,\n })\n .from(skills)\n .where(\n and(\n eq(skills.deleted, false),\n eq(skills.duplicate, false),\n ),\n )\n .orderBy(desc(skills.updatedAt))\n .limit(limit)\n .offset(offset)\n .all();\n\n const urls = results.map((skill) => {\n const date = skill.updatedAt ? new Date(skill.updatedAt) : new Date();\n const isValidDate = !isNaN(date.getTime());\n\n const daysSinceUpdate = isValidDate\n ? (Date.now() - date.getTime()) / (1000 * 60 * 60 * 24)\n : Infinity;\n\n return {\n loc: `${BASE_URL}/skills/${skill.namespace}`,\n lastmod: date.toISOString()?.split(\"T\")[0],\n priority: 0.8,\n changefreq: daysSinceUpdate < 7 ? \"daily\" : \"weekly\",\n };\n });\n\n return { urls, total, limit, offset };\n },\n SITEMAP_CACHE_TTL_MS,\n );\n\n return c.json(result);\n}\n\n/**\n * GET /api/sitemap/plugins\n * Returns paginated plugins for sitemap generation (cached for 1 day)\n * Query params:\n * - limit: max 50000 (default 50000)\n * - offset: default 0\n */\nexport async function getPluginsSitemap(c: Context) {\n const limit = Math.min(\n parseInt(c.req.query(\"limit\") || String(DEFAULT_LIMIT)),\n MAX_SITEMAP_LIMIT,\n );\n const offset = parseInt(c.req.query(\"offset\") || \"0\");\n\n const result = await getCached(\n getSitemapPluginsCacheKey(limit, offset),\n async () => {\n const total = await getPluginsCount();\n\n const results = await db\n .select({\n namespace: plugins.namespace,\n name: plugins.name,\n updatedAt: plugins.updatedAt,\n })\n .from(plugins)\n .orderBy(desc(plugins.updatedAt))\n .limit(limit)\n .offset(offset)\n .all();\n\n const urls = results.map((plugin) => ({\n loc: `${BASE_URL}/plugins/${plugin.namespace}/${plugin.name}`,\n lastmod: plugin.updatedAt?.split(\"T\")[0],\n }));\n\n return { urls, total, limit, offset };\n },\n SITEMAP_CACHE_TTL_MS,\n );\n\n return c.json(result);\n}",[],[308,328],"0cfb8c44-0a11-11f1-8914-42dde27851f2","search.ts","019c5f00-3df8-778f-befc-c115951d23c3",["D",1771120376884],"/**\n * Plugin search endpoint\n * Search and filter plugins by query, category, etc.\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, asc, desc, eq, SQL, sql } from \"npm:drizzle-orm@0.44.2\";\nimport { db } from \"../database/db.ts\";\nimport { plugins } from \"../database/schema.ts\";\nimport { getPluginsCount } from \"../clients/cache.ts\";\nimport { sqlite } from \"../clients/sqlite.ts\";\nimport { sanitizeFtsQuery } from \"../utils.ts\";\n\nconst ALLOWED_SORT_COLUMNS = new Set([\n \"downloads\",\n \"stars\",\n]);\n\nfunction getPluginColumn(field: string) {\n switch (field) {\n case \"downloads\":\n return plugins.downloads;\n case \"stars\":\n return plugins.stars;\n default:\n return null;\n }\n}\n\nexport function getSkillName(path: string): string {\n const prefix = \"./skills/\";\n return path.startsWith(prefix) ? path.slice(prefix.length) : path;\n}\n\n/**\n * GET /api/search?q=supabase&category=database&limit=20&offset=0\n * Search plugins with optional filters\n */\nexport async function searchPlugins(c: Context) {\n const query = c.req.query(\"q\") || \"\";\n const category = c.req.query(\"category\");\n const hasSkills = c.req.query(\"hasSkills\");\n const limit = Math.min(parseInt(c.req.query(\"limit\") || \"20\"), 100); // Max 100\n const offset = parseInt(c.req.query(\"offset\") || \"0\");\n const orderByParam = c.req.query(\"orderBy\");\n const orderParam = c.req.query(\"order\") ?? \"desc\";\n\n if (orderByParam && !ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n return c.json({ error: \"Invalid orderBy parameter\" }, 400);\n }\n\n const isUnfilteredQuery = !query && !category && hasSkills !== \"true\";\n const ftsQuery = sanitizeFtsQuery(query);\n\n // Use FTS for text queries\n if (ftsQuery) {\n return searchPluginsWithFTS(\n c,\n {\n query: ftsQuery,\n category,\n hasSkills,\n limit,\n offset,\n orderByParam,\n orderParam,\n },\n );\n }\n\n // Build where conditions for non-query filters\n const conditions: SQL[] = [];\n\n if (category) {\n conditions.push(eq(plugins.category, category));\n }\n\n if (hasSkills === \"true\") {\n const hasSkillsFilter =\n sql`json_type(${plugins.skills}) = 'array' AND json_array_length(${plugins.skills}) >= 1`;\n conditions.push(hasSkillsFilter);\n }\n\n const finalWhere = conditions.length === 0\n ? undefined\n : conditions.length === 1\n ? conditions[0]\n : and(...conditions);\n\n // Get total count\n let total: number;\n if (isUnfilteredQuery) {\n total = await getPluginsCount();\n } else {\n total = await (async () => {\n const countQuery = db.select({ count: sql`count(*)` }).from(\n plugins,\n );\n const [{ count }] = finalWhere\n ? await countQuery.where(finalWhere).all()\n : await countQuery.all();\n return count;\n })();\n }\n\n // Build data query - use partial select for unfiltered queries\n const baseSelect = isUnfilteredQuery\n ? db.select({\n id: plugins.id,\n name: plugins.name,\n namespace: plugins.namespace,\n gitUrl: plugins.gitUrl,\n description: plugins.description,\n version: plugins.version,\n author: plugins.author,\n keywords: plugins.keywords,\n skills: plugins.skills,\n category: plugins.category,\n stars: plugins.stars,\n verified: plugins.verified,\n downloads: plugins.downloads,\n metadata: plugins.metadata,\n createdAt: plugins.createdAt,\n updatedAt: plugins.updatedAt,\n }).from(plugins)\n : db.select().from(plugins);\n\n const withWhere = finalWhere ? baseSelect.where(finalWhere) : baseSelect;\n\n let orderedQuery;\n if (orderByParam && ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n const column = getPluginColumn(orderByParam);\n const sortDirection = orderParam === \"asc\" ? asc : desc;\n\n if (!column) {\n throw new Error(`Invalid orderBy param: ${orderByParam}`);\n }\n orderedQuery = withWhere.orderBy(sortDirection(column));\n } else {\n orderedQuery = withWhere.orderBy(\n desc(plugins.downloads),\n desc(plugins.stars),\n );\n }\n\n const results = await orderedQuery.limit(limit).offset(offset).all();\n const parseSkills = (skills: string) => {\n const parsedSkills = JSON.parse(skills);\n if (Array.isArray(parsedSkills) && parsedSkills.length > 0) {\n return parsedSkills?.map(getSkillName);\n }\n\n return parsedSkills;\n };\n\n // Format results\n const formattedPlugins = results.map((plugin) => ({\n id: plugin.id,\n name: plugin.name,\n namespace: plugin.namespace,\n gitUrl: plugin.gitUrl,\n description: plugin.description,\n version: plugin.version,\n author: plugin.author,\n keywords: plugin.keywords ? JSON.parse(plugin.keywords) : [],\n skills: plugin.skills ? parseSkills(plugin.skills) : [],\n category: plugin.category,\n stars: plugin.stars,\n verified: Boolean(plugin.verified),\n downloads: plugin.downloads,\n metadata: plugin.metadata ? JSON.parse(plugin.metadata) : {},\n createdAt: plugin.createdAt,\n updatedAt: plugin.updatedAt,\n }));\n\n return c.json({\n plugins: formattedPlugins,\n total,\n limit,\n offset,\n });\n}\n\n/**\n * FTS5 full-text search for plugins\n * Falls back to LIKE-based search if FTS fails\n */\nasync function searchPluginsWithFTS(\n c: Context,\n {\n query: ftsQuery,\n category,\n hasSkills,\n limit,\n offset,\n orderByParam,\n orderParam,\n }: {\n query: string;\n category: string | undefined;\n hasSkills: string | undefined;\n limit: number;\n offset: number;\n orderByParam: string | undefined;\n orderParam: string;\n },\n) {\n if (!ftsQuery) {\n // If sanitization removed everything, fall back to listing without query\n return searchPlugins(c);\n }\n\n try {\n // Build additional WHERE clauses for category and hasSkills\n let additionalWhere = \"\";\n const args: (string | number)[] = [ftsQuery];\n\n if (category) {\n additionalWhere += \" AND plugins.category = ?\";\n args.push(category);\n }\n\n if (hasSkills === \"true\") {\n additionalWhere +=\n \" AND json_type(plugins.skills) = 'array' AND json_array_length(plugins.skills) >= 1\";\n }\n\n const total = await (async () => {\n const countSql = category || hasSkills === \"true\"\n ? `SELECT COUNT(*) FROM plugins_fts \n JOIN plugins ON plugins.rowid = plugins_fts.rowid\n WHERE plugins_fts MATCH ?${additionalWhere}`\n : `SELECT COUNT(*) FROM plugins_fts WHERE plugins_fts MATCH ?`;\n\n const countArgs = category ?? hasSkills === \"true\" ? args : [ftsQuery];\n const countResult = await sqlite.execute({\n sql: countSql,\n args: countArgs,\n });\n return Number(countResult.rows[0]?.[0] ?? 0);\n })();\n\n // Build ORDER BY clause\n let orderClause: string;\n if (orderByParam && ALLOWED_SORT_COLUMNS.has(orderByParam)) {\n const direction = orderParam === \"asc\" ? \"ASC\" : \"DESC\";\n const column = orderByParam === \"downloads\"\n ? \"plugins.downloads\"\n : \"plugins.stars\";\n orderClause = `${column} ${direction}`;\n } else {\n // Combined score: BM25 relevance + popularity boost\n orderClause =\n `(bm25(plugins_fts, 0, 10.0, 5.0, 1.0, 1.0) - (ln(plugins.downloads + 1) * 2.0) - (ln(plugins.stars + 1) * 1.0)) ASC`;\n }\n\n // Execute FTS5 search\n const searchArgs = [...args, limit, offset];\n const results = await sqlite.execute({\n sql: `\n SELECT\n plugins.id,\n plugins.name,\n plugins.namespace,\n plugins.git_url,\n plugins.description,\n plugins.version,\n plugins.author,\n plugins.keywords,\n plugins.skills,\n plugins.category,\n plugins.stars,\n plugins.verified,\n plugins.downloads,\n plugins.metadata,\n plugins.created_at,\n plugins.updated_at\n FROM plugins_fts\n JOIN plugins ON plugins.rowid = plugins_fts.rowid\n WHERE plugins_fts MATCH ?${additionalWhere}\n ORDER BY ${orderClause}\n LIMIT ? OFFSET ?\n `,\n args: searchArgs,\n });\n\n const parseSkills = (skills: string | null) => {\n if (!skills) return [];\n const parsedSkills = JSON.parse(skills);\n if (Array.isArray(parsedSkills) && parsedSkills.length > 0) {\n return parsedSkills?.map(getSkillName);\n }\n return parsedSkills;\n };\n\n const formattedPlugins = results.rows.map((row: any) => ({\n id: row[0],\n name: row[1],\n namespace: row[2],\n gitUrl: row[3],\n description: row[4],\n version: row[5],\n author: row[6],\n keywords: row[7] ? JSON.parse(row[7]) : [],\n skills: parseSkills(row[8]),\n category: row[9],\n stars: row[10],\n verified: Boolean(row[11]),\n downloads: row[12],\n metadata: row[13] ? JSON.parse(row[13]) : {},\n createdAt: row[14],\n updatedAt: row[15],\n }));\n\n return c.json({\n plugins: formattedPlugins,\n total,\n limit,\n offset,\n });\n } catch (error) {\n // Log and fall back to LIKE-based search\n console.warn(\"FTS5 search failed\", error);\n return c.json({\n error: \"Search temporarily unavailable. Please try again.\",\n details: String(error),\n }, 503);\n }\n}",[],[308,335],"0cfb7bf0-0a11-11f1-8914-42dde27851f2","resolve.ts","019c5f00-3df8-778f-befc-be1f3138ea05",["D",1771120376884],"/**\n * Plugin resolution endpoint\n * Resolves @namespace/name to git URL\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, eq, sql } from \"npm:drizzle-orm@0.44.2\";\nimport { db } from \"../database/db.ts\";\nimport { plugins, pluginStats } from \"../database/schema.ts\";\nimport { getLogger } from \"https://esm.town/v/kamalnrf/axiom-logger/hono.ts\";\n\n/**\n * GET /api/resolve/:owner/:marketplace/:plugin\n * Resolves plugin identifier to git URL and metadata\n *\n * Examples:\n * /api/resolve/davila7/claude-code-templates/supabase-toolkit\n * /api/resolve/anthropics/claude-code-plugins/agent-sdk-dev\n */\nexport async function resolvePlugin(c: Context) {\n const logger = getLogger(c);\n const { owner, marketplace, plugin: pluginName } = c.req.param();\n logger.info(\"Resolved Fields\", { owner, marketplace, pluginName });\n\n if (!owner || !marketplace || !pluginName) {\n return c.json({ error: \"Missing owner, marketplace, or plugin name\" }, 400);\n }\n\n const ownerName = owner.startsWith(\"@\") ? owner : `@${owner}`;\n\n const namespace = `${ownerName}/${marketplace}`;\n logger.info(\"Namespace:\", { namespace });\n\n // Query database\n const plugin = await db\n .select()\n .from(plugins)\n .where(and(eq(plugins.namespace, namespace), eq(plugins.name, pluginName)))\n .get();\n\n if (!plugin) {\n return c.json({\n error: `Plugin ${owner}/${marketplace}/${pluginName} not found`,\n }, 404);\n }\n\n // Increment download count\n try {\n await incrementDownloads(plugin.id);\n } catch (error) {\n logger.error(\"Failed to increment downloads:\", { error });\n // Don't fail the request if stats update fails\n }\n\n // Parse JSON fields\n const keywords = plugin.keywords ? JSON.parse(plugin.keywords) : [];\n const metadata = plugin.metadata ? JSON.parse(plugin.metadata) : {};\n\n return c.json({\n name: plugin.name,\n namespace: plugin.namespace,\n gitUrl: plugin.gitUrl,\n description: plugin.description,\n version: plugin.version,\n author: plugin.author,\n keywords,\n category: plugin.category,\n stars: plugin.stars,\n verified: Boolean(plugin.verified),\n downloads: plugin.downloads,\n metadata,\n });\n}\n\n/**\n * Increment download statistics for a plugin\n * Uses upsert to avoid extra SELECT query (saves 1 row read per download)\n */\nasync function incrementDownloads(pluginId: string) {\n const now = new Date().toISOString();\n\n // Update total downloads in plugins table\n await db\n .update(plugins)\n .set({ downloads: sql`${plugins.downloads} + 1` })\n .where(eq(plugins.id, pluginId));\n\n // Upsert stats\n await db\n .insert(pluginStats)\n .values({\n pluginId,\n downloadsTotal: 1,\n downloadsWeek: 1,\n downloadsMonth: 1,\n lastDownloadedAt: now,\n })\n .onConflictDoUpdate({\n target: pluginStats.pluginId,\n set: {\n downloadsTotal: sql`${pluginStats.downloadsTotal} + 1`,\n downloadsWeek: sql`${pluginStats.downloadsWeek} + 1`,\n downloadsMonth: sql`${pluginStats.downloadsMonth} + 1`,\n lastDownloadedAt: now,\n },\n });\n}\n",[],[308,342],"0cfb757e-0a11-11f1-8914-42dde27851f2","plugins.ts","019c5f00-3df8-778f-befc-bb36d5e1e26c",["D",1771120376884],"/**\n * Plugin stats endpoint\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, eq } from \"npm:drizzle-orm@0.44.2\";\nimport { db } from \"../database/db.ts\";\nimport { plugins, pluginStats } from \"../database/schema.ts\";\n\n/**\n * GET /api/plugins/:owner/:marketplace/:plugin/stats\n * Get plugin download statistics\n */\nexport async function getPluginStats(c: Context) {\n const { owner, marketplace, plugin: pluginName } = c.req.param();\n const namespace = `${owner}/${marketplace}`;\n\n const result = await db\n .select({\n plugin: plugins,\n stats: pluginStats,\n })\n .from(plugins)\n .leftJoin(pluginStats, eq(plugins.id, pluginStats.pluginId))\n .where(and(eq(plugins.namespace, namespace), eq(plugins.name, pluginName)))\n .get();\n\n if (!result || !result.plugin) {\n return c.json({\n error: `Plugin ${owner}/${marketplace}/${pluginName} not found`,\n }, 404);\n }\n\n return c.json({\n downloads: {\n total: result.stats?.downloadsTotal || 0,\n week: result.stats?.downloadsWeek || 0,\n month: result.stats?.downloadsMonth || 0,\n },\n lastDownloaded: result.stats?.lastDownloadedAt || null,\n });\n}",[],[308,349],"0cfb62fa-0a11-11f1-8914-42dde27851f2","index-repo-skills.ts","019c5f00-3df8-778f-befc-b77f9fb7fc43",["D",1771120376884],"/**\n * API endpoint to index skills from a specific GitHub repository\n * POST /api/skills/index\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { GitHubClient } from \"../clients/github.ts\";\nimport { indexSkillFiles } from \"../lib/skills/index-skill-files.ts\";\nimport { discoverSkillsInRepo, DiscoverSkillsResult } from \"../lib/skills/discover.ts\";\nimport { getLogger } from \"https://esm.town/v/kamalnrf/axiom-logger/hono.ts\";\n\nconst github = new GitHubClient();\n\n/**\n * Normalize repo input to \"owner/repo\" format\n * Accepts: \"owner/repo\", \"https://github.com/owner/repo\", \"github.com/owner/repo\"\n */\nfunction normalizeRepoInput(input: string): string | null {\n if (!input) return null;\n\n // Already in owner/repo format\n if (/^[\\w.-]+\\/[\\w.-]+$/.test(input)) {\n return input;\n }\n\n // URL format\n const urlMatch = input.match(\n /(?:https?:\\/\\/)?(?:www\\.)?github\\.com\\/([\\w.-]+)\\/([\\w.-]+)/,\n );\n if (urlMatch) {\n return `${urlMatch[1]}/${urlMatch[2]}`;\n }\n\n return null;\n}\n\nexport async function indexRepoSkills(c: Context) {\n const startTime = Date.now();\n const log = getLogger(c);\n\n try {\n const body = await c.req.json();\n const { repo, dryRun = false, maxFiles = 200 } = body;\n\n // Validate input\n const repoFullName = normalizeRepoInput(repo);\n if (!repoFullName) {\n return c.json(\n {\n ok: false,\n error: \"Invalid repo format. Expected 'owner/repo' or GitHub URL.\",\n },\n 400,\n );\n }\n\n log.info(`Indexing skills from repo: ${repoFullName}`);\n\n // Discover all SKILL.md files in the repo\n let discovered: DiscoverSkillsResult;\n try {\n discovered = await discoverSkillsInRepo(github, repoFullName);\n } catch (error) {\n const message = error instanceof Error ? error.message : String(error);\n if (message.includes(\"404\")) {\n return c.json(\n { ok: false, error: `Repository not found: ${repoFullName}` },\n 404,\n );\n }\n throw error;\n }\n\n if (discovered.skills.length === 0) {\n return c.json({\n ok: true,\n repo: repoFullName,\n defaultBranch: discovered.defaultBranch,\n pathsFound: 0,\n result: {\n found: 0,\n indexed: 0,\n unchanged: 0,\n invalid: 0,\n failed: 0,\n skipped: 0,\n duplicatesMarked: 0,\n skills: [],\n },\n durationMs: Date.now() - startTime,\n });\n }\n\n // Apply maxFiles limit\n const refs = discovered.skills.slice(0, maxFiles);\n if (discovered.skills.length > maxFiles) {\n log.warn(\n `Found ${discovered.skills.length} SKILL.md files, limiting to ${maxFiles}`,\n );\n }\n\n // Index the skills\n const result = await indexSkillFiles(refs, { github, log, dryRun });\n\n return c.json({\n ok: true,\n repo: repoFullName,\n defaultBranch: discovered.defaultBranch,\n pathsFound: discovered.skills.length,\n truncated: discovered.truncated,\n limitApplied: discovered.skills.length > maxFiles ? maxFiles : undefined,\n result,\n dryRun,\n durationMs: Date.now() - startTime,\n });\n } catch (error) {\n log.error(\"Failed to index repo skills\", { error: String(error) });\n return c.json(\n {\n ok: false,\n error: error instanceof Error ? error.message : \"Unknown error\",\n },\n 500,\n );\n }\n};\n",[],[308,356],"0cfb602a-0a11-11f1-8914-42dde27851f2","v2","019c5f00-3df8-778f-befc-b256d90d485e",["D",1771120376884],[368],[308,363],{"_36":369,"_13":342,"_62":370,"_64":364,"_66":67,"_68":34,"_69":70,"_38":371,"_17":32,"_72":97,"_74":372,"_75":51,"_76":-5,"_99":373,"_18":374},"0cfb0706-0a11-11f1-8914-42dde27851f2","019c5f00-3df8-778f-befc-aca7bc69ecb6",["D",1771120376884],"/**\n * Unified skill resolution API endpoint\n * POST /api/v2/skills/resolve\n *\n * Handles different input formats and returns installable skills with sourceUrl:\n * - owner → list skills from all repos by that owner\n * - owner/repo → list skills in that repo\n * - owner/repo/skill → resolve specific skill\n */\nimport type { Context } from \"npm:hono@4.9\";\nimport { and, eq, like } from \"npm:drizzle-orm@0.44.2\";\nimport { db } from \"../../database/db.ts\";\nimport { skills } from \"../../database/schema.ts\";\nimport { GitHubClient } from \"../../clients/github.ts\";\nimport { discoverSkillsInRepo } from \"../../lib/skills/discover.ts\";\nimport { getLogger } from \"https://esm.town/v/kamalnrf/axiom-logger/hono.ts\";\n\nconst github = new GitHubClient();\n\n// Types\ntype Status = \"success\" | \"error\";\ntype TargetKind = \"owner\" | \"repo\" | \"skill\";\n\ninterface ResolveRequest {\n target: string;\n limit?: number;\n offset?: number;\n}\n\ninterface ResolvedSkill {\n namespace: string;\n name: string;\n relDir: string;\n sourceUrl: string;\n}\n\ninterface ResolveResponse {\n status: Status;\n query: {\n target: string;\n kind: TargetKind;\n normalized: string;\n };\n skills: ResolvedSkill[];\n page: {\n limit: number;\n offset: number;\n hasMore: boolean;\n total: number;\n };\n error?: string;\n}\n\n/**\n * Parse and normalize the target input\n * Returns { kind, owner, repo?, skill?, normalized }\n */\nfunction parseTarget(input: string): {\n kind: TargetKind;\n owner: string;\n repo?: string;\n skill?: string;\n normalized: string;\n} | null {\n if (!input || typeof input !== \"string\") return null;\n\n let cleaned = input.trim();\n\n // Strip GitHub URL prefixes\n cleaned = cleaned\n .replace(/^https?:\\/\\/(www\\.)?github\\.com\\//, \"\")\n .replace(/^github\\.com\\//, \"\")\n .replace(/^git@github\\.com:/, \"\")\n .replace(/\\.git$/, \"\");\n\n // Strip leading @\n cleaned = cleaned.replace(/^@/, \"\");\n\n // Remove any tree/blob path segments (e.g., /tree/main/skills/foo)\n const treeMatch = cleaned.match(\n /^([\\w.-]+)\\/([\\w.-]+)\\/(?:tree|blob)\\/[^/]+\\/(.+)$/,\n );\n if (treeMatch) {\n const [, owner, repo, subpath] = treeMatch;\n // Try to extract skill name from path like \"skills/foo\" or \"skills/foo/SKILL.md\"\n const skillMatch = subpath?.match(\n /(?:skills?\\/)?([\\w.-]+)(?:\\/SKILL\\.md)?$/,\n );\n if (skillMatch) {\n return {\n kind: \"skill\",\n owner: owner!,\n repo: repo!,\n skill: skillMatch[1],\n normalized: `${owner}/${repo}/${skillMatch[1]}`,\n };\n }\n }\n\n const parts = cleaned.split(\"/\").filter(Boolean);\n\n if (parts.length === 1) {\n return {\n kind: \"owner\",\n owner: parts[0]!,\n normalized: parts[0]!,\n };\n }\n\n if (parts.length === 2) {\n return {\n kind: \"repo\",\n owner: parts[0]!,\n repo: parts[1]!,\n normalized: `${parts[0]}/${parts[1]}`,\n };\n }\n\n if (parts.length >= 3) {\n // Take first 3 segments as owner/repo/skill\n return {\n kind: \"skill\",\n owner: parts[0]!,\n repo: parts[1]!,\n skill: parts[2]!,\n normalized: `${parts[0]}/${parts[1]}/${parts[2]}`,\n };\n }\n\n return null;\n}\n\n/**\n * Query skills from DB by owner\n */\nasync function querySkillsByOwner(\n owner: string,\n limit: number,\n offset: number,\n): Promise<{ skills: ResolvedSkill[]; total: number }> {\n const results = await db\n .select()\n .from(skills)\n .where(\n and(\n eq(skills.author, owner),\n eq(skills.deleted, false),\n eq(skills.duplicate, false),\n ),\n )\n .limit(limit)\n .offset(offset)\n .all()\n\n return {\n skills: results.map(mapDbSkillToResolved),\n total: results.length,\n };\n}\n\n/**\n * Query a specific skill from DB\n */\nasync function querySkillByNamespace(\n namespace: string,\n): Promise {\n const result = await db\n .select()\n .from(skills)\n .where(\n and(\n eq(skills.namespace, namespace),\n eq(skills.deleted, false),\n ),\n )\n .get();\n\n return result ? mapDbSkillToResolved(result) : null;\n}\n\n/**\n * Map DB skill row to API response format\n */\nfunction mapDbSkillToResolved(\n skill: typeof skills.$inferSelect,\n): ResolvedSkill {\n const metadata = skill.metadata ? JSON.parse(skill.metadata) : {};\n\n return {\n namespace: skill.namespace,\n name: skill.name,\n relDir: metadata.directoryPath ?? \"\",\n sourceUrl: skill.sourceUrl,\n };\n}\n\n/**\n * Discover skills in a repo (no indexing, just discovery via GitHub trees API)\n * Returns skills with constructed sourceUrls for installation\n */\nasync function discoverSkillsInRepoWithSourceUrls(\n owner: string,\n repo: string,\n): Promise<{ skills: ResolvedSkill[]; defaultBranch: string }> {\n const repoFullName = `${owner}/${repo}`;\n\n // Discover skills in repo using GitHub trees API\n const discovered = await discoverSkillsInRepo(github, repoFullName);\n\n if (discovered.skills.length === 0) {\n return { skills: [], defaultBranch: discovered.defaultBranch };\n }\n\n // Map discovered skills to response format\n // Each skill has: repoFullName, repoHtmlUrl, path (e.g., \"skills/foo/SKILL.md\")\n const resolvedSkills: ResolvedSkill[] = discovered.skills.map((ref) => {\n // Get skill directory from SKILL.md path (remove /SKILL.md suffix)\n const skillDir = ref.path.replace(/\\/SKILL\\.md$/, \"\");\n // Extract skill name from the directory path (last segment)\n const skillName = skillDir.split(\"/\").pop() ?? \"\";\n // Construct sourceUrl: GitHub tree URL pointing to skill directory\n const sourceUrl =\n `${ref.repoHtmlUrl}/tree/${discovered.defaultBranch}/${skillDir}`;\n // Namespace format: @owner/repo/skillName\n const namespace = `@${owner}/${repo}/${skillName}`;\n\n return {\n namespace,\n name: skillName,\n relDir: skillDir,\n sourceUrl,\n };\n });\n\n return { skills: resolvedSkills, defaultBranch: discovered.defaultBranch };\n}\n\n/**\n * Main resolve endpoint handler\n */\nexport async function resolveSkills(c: Context) {\n const log = getLogger(c);\n\n try {\n const body = (await c.req.json()) as ResolveRequest;\n const { target, limit = 50, offset = 0 } = body;\n\n // Validate and parse target\n const parsed = parseTarget(target);\n if (!parsed) {\n return c.json(\n {\n status: \"error\",\n error:\n \"Invalid target format. Expected: owner, owner/repo, or owner/repo/skill\",\n },\n 400,\n );\n }\n\n const { kind, owner, repo, skill, normalized } = parsed;\n\n // Clamp limits\n const safeLimit = Math.min(Math.max(1, limit), 100);\n const safeOffset = Math.max(0, offset);\n\n log.info(\"Resolving skills\", { target, kind, normalized });\n\n // Handle by target kind\n switch (kind) {\n case \"owner\": {\n // Owner queries are registry-only (never trigger indexing)\n const result = await querySkillsByOwner(owner, safeLimit, safeOffset);\n\n return c.json({\n status: \"success\",\n query: { target, kind, normalized },\n skills: result.skills,\n page: {\n limit: safeLimit,\n offset: safeOffset,\n hasMore: safeOffset + result.skills.length < result.total,\n total: result.total,\n },\n } as ResolveResponse);\n }\n\n case \"repo\": {\n try {\n // Discover skills via GitHub trees API (fast, no indexing)\n const { skills: discoveredSkills } =\n await discoverSkillsInRepoWithSourceUrls(\n owner,\n repo!,\n );\n\n // Apply pagination\n const paginatedSkills = discoveredSkills.slice(\n safeOffset,\n safeOffset + safeLimit,\n );\n\n return c.json({\n status: \"success\",\n query: { target, kind, normalized },\n skills: paginatedSkills,\n page: {\n limit: safeLimit,\n offset: safeOffset,\n hasMore:\n safeOffset + paginatedSkills.length < discoveredSkills.length,\n total: discoveredSkills.length,\n },\n } as ResolveResponse);\n } catch (error) {\n const message = error instanceof Error\n ? error.message\n : String(error);\n if (message.includes(\"404\")) {\n return c.json(\n {\n status: \"error\",\n query: { target, kind, normalized },\n error: `Repository not found: ${owner}/${repo}`,\n } as Partial,\n 404,\n );\n }\n throw error;\n }\n }\n\n case \"skill\": {\n const namespace = `@${owner}/${repo}/${skill}`;\n\n // Try to find in DB first\n let resolvedSkill = await querySkillByNamespace(namespace);\n\n if (!resolvedSkill) {\n // Not in DB, discover from repo to get sourceUrl\n try {\n const { skills: discoveredSkills } =\n await discoverSkillsInRepoWithSourceUrls(\n owner,\n repo!,\n );\n // Find the specific skill by name\n resolvedSkill = discoveredSkills.find((s) => s.name === skill) ??\n null;\n\n return c.json({\n status: \"success\",\n query: { target, kind, normalized },\n skills: resolvedSkill ? [resolvedSkill] : [],\n page: {\n limit: 1,\n offset: 0,\n hasMore: false,\n total: resolvedSkill ? 1 : 0,\n },\n } as ResolveResponse);\n } catch (error) {\n const message = error instanceof Error\n ? error.message\n : String(error);\n if (message.includes(\"404\")) {\n return c.json(\n {\n status: \"error\",\n query: { target, kind, normalized },\n error: `Repository not found: ${owner}/${repo}`,\n } as Partial,\n 404,\n );\n }\n throw error;\n }\n }\n\n return c.json({\n status: \"success\",\n query: { target, kind, normalized },\n skills: resolvedSkill ? [resolvedSkill] : [],\n page: {\n limit: 1,\n offset: 0,\n hasMore: false,\n total: resolvedSkill ? 1 : 0,\n },\n } as ResolveResponse);\n }\n }\n } catch (error) {\n log.error(\"Failed to resolve skills\", { error: String(error) });\n return c.json(\n {\n status: \"error\",\n error: error instanceof Error ? error.message : \"Unknown error\",\n } as Partial,\n 500,\n );\n }\n}",[],[308,363,342],"0cba5710-0a11-11f1-8914-42dde27851f2",["D",1771120376884],"legacyEnvVars","httpPrivacy","description","A lightweight registry for Agent Skills and Plugins","imageUrl","userId","019c5eed-607e-7487-b78d-e19639f56f0a","pinnedFileId","pinPreview","pinMigrated","screenshotVersion","likeCount","019c5eed-607e-7487-b78d-e6e09525a366","tursoName","0cba57100a1111f1891442dde27851f2","tursoHostname","0cba57100a1111f1891442dde27851f2-val-town.aws-us-east-2.turso.io","tursoMeasuredAt",["D",1789604441792],"pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_11":12,"_422":418},"org",{"_11":12,"_416":-5,"_417":418,"_382":383,"_419":420,"_421":51},"branch",{"_17":34,"_13":16,"_36":67,"_42":375,"_45":37,"_46":412,"_413":53,"_414":43,"_415":57,"_58":14,"_54":12},665,"forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","displayName","avatar","https://img.clerk.com/eyJ0eXBlIjoicHJveHkiLCJzcmMiOiJodHRwczovL2ltYWdlcy5jbGVyay5kZXYvb2F1dGhfZ2l0aHViL2ltZ18zOWdZRTJDRUp1bHFRaVc4dEJzZ1VMSlNqTHAifQ","tier","free","canRestrictVals","profileImageUrl","adoptOrgId",{"_36":375,"_13":14,"_38":575,"_49":-5,"_377":51,"_52":53,"_378":53,"_379":380,"_381":-5,"_382":383,"_384":104,"_385":51,"_386":-5,"_387":-5,"_388":32,"_56":389,"_390":391,"_392":393,"_394":576,"_396":-5,"_397":398,"_399":398,"_400":401,"_402":401,"_403":401,"_404":398,"_405":398,"_406":577,"_408":578,"_410":579,"_50":51},[461,470,474,481,485,492,493,494,495,496,497,498],"files",[461,462,463,464,465,466,467,468,469,470,471,472,473,474,475,476,477,478,479,480,481,482,483,484,485,486,487,488,489,490,491,492,493,494,495,496,497,498],"counts",{"_451":32,"_452":34,"_453":32,"_35":454,"_455":456},[449],{"_36":78,"_13":19,"_62":65,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":79,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_18":448},"upstreamProject",{"_36":43,"_13":14,"_379":380,"_381":-5,"_52":53,"_406":446},"upstreamBranch",{"_36":37,"_38":444,"_40":445,"_13":16,"_42":43,"_17":44,"_45":-5,"_46":-5,"_47":48,"_49":-5,"_50":51,"_52":53,"_54":55,"_56":57,"_58":14},"emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},["D",1760263652191],["D",1778300946443],{"_11":55,"_422":447},"https://img.clerk.com/eyJ0eXBlIjoicHJveHkiLCJzcmMiOiJodHRwczovL2ltYWdlcy5jbGVyay5kZXYvb2F1dGhfZ2l0aHViL2ltZ18yaVpXT3l1d2hWSFdtWTNoZFIwZ0lFUUhOd2oifQ",[19],{"_36":60,"_13":61,"_62":63,"_64":65,"_66":67,"_68":34,"_69":70,"_38":71,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_18":450},[19,61],"pulls","branches","forks",{"_457":459,"_460":459},"connections",{"_457":32,"_458":32},"count","unfilledCount",10,"emptyCount",{"_36":307,"_13":308,"_62":309,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":310,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":573,"_18":574},{"_36":362,"_13":363,"_62":364,"_64":309,"_66":67,"_68":34,"_69":70,"_38":365,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":571,"_18":572},{"_36":369,"_13":342,"_62":370,"_64":364,"_66":67,"_68":34,"_69":70,"_38":371,"_17":32,"_72":97,"_74":372,"_75":51,"_76":-5,"_99":569,"_18":570},{"_36":355,"_13":356,"_62":357,"_64":309,"_66":67,"_68":34,"_69":70,"_38":358,"_17":32,"_72":97,"_74":359,"_75":51,"_76":-5,"_99":567,"_18":568},{"_36":348,"_13":349,"_62":350,"_64":309,"_66":67,"_68":34,"_69":70,"_38":351,"_17":32,"_72":97,"_74":352,"_75":51,"_76":-5,"_99":565,"_18":566},{"_36":341,"_13":342,"_62":343,"_64":309,"_66":67,"_68":34,"_69":70,"_38":344,"_17":32,"_72":97,"_74":345,"_75":51,"_76":-5,"_99":563,"_18":564},{"_36":334,"_13":335,"_62":336,"_64":309,"_66":67,"_68":34,"_69":70,"_38":337,"_17":32,"_72":97,"_74":338,"_75":51,"_76":-5,"_99":561,"_18":562},{"_36":327,"_13":328,"_62":329,"_64":309,"_66":67,"_68":34,"_69":70,"_38":330,"_17":32,"_72":97,"_74":331,"_75":51,"_76":-5,"_99":559,"_18":560},{"_36":320,"_13":321,"_62":322,"_64":309,"_66":67,"_68":34,"_69":70,"_38":323,"_17":32,"_72":97,"_74":324,"_75":51,"_76":-5,"_99":557,"_18":558},{"_36":277,"_13":278,"_62":279,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":280,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":555,"_18":556},{"_36":300,"_13":301,"_62":302,"_64":279,"_66":67,"_68":34,"_69":70,"_38":303,"_17":32,"_72":97,"_74":304,"_75":51,"_76":-5,"_99":553,"_18":554},{"_36":293,"_13":294,"_62":295,"_64":279,"_66":67,"_68":34,"_69":70,"_38":296,"_17":32,"_72":97,"_74":297,"_75":51,"_76":-5,"_99":551,"_18":552},{"_36":286,"_13":287,"_62":288,"_64":279,"_66":67,"_68":34,"_69":70,"_38":289,"_17":32,"_72":97,"_74":290,"_75":51,"_76":-5,"_99":549,"_18":550},{"_36":222,"_13":223,"_62":224,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":225,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":547,"_18":548},{"_36":270,"_13":271,"_62":272,"_64":224,"_66":67,"_68":34,"_69":70,"_38":273,"_17":32,"_72":97,"_74":274,"_75":51,"_76":-5,"_99":545,"_18":546},{"_36":263,"_13":264,"_62":265,"_64":224,"_66":67,"_68":34,"_69":70,"_38":266,"_17":32,"_72":97,"_74":267,"_75":51,"_76":-5,"_99":543,"_18":544},{"_36":256,"_13":257,"_62":258,"_64":224,"_66":67,"_68":34,"_69":70,"_38":259,"_17":32,"_72":238,"_74":260,"_75":51,"_76":-5,"_99":541,"_18":542},{"_36":249,"_13":250,"_62":251,"_64":224,"_66":67,"_68":34,"_69":70,"_38":252,"_17":32,"_72":238,"_74":253,"_75":51,"_76":-5,"_99":539,"_18":540},{"_36":242,"_13":243,"_62":244,"_64":224,"_66":67,"_68":34,"_69":70,"_38":245,"_17":32,"_72":238,"_74":246,"_75":51,"_76":-5,"_99":537,"_18":538},{"_36":234,"_13":235,"_62":236,"_64":224,"_66":67,"_68":34,"_69":70,"_38":237,"_17":32,"_72":238,"_74":239,"_75":51,"_76":-5,"_99":535,"_18":536},{"_36":192,"_13":193,"_62":194,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":195,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":533,"_18":534},{"_36":215,"_13":216,"_62":217,"_64":194,"_66":67,"_68":34,"_69":70,"_38":218,"_17":32,"_72":97,"_74":219,"_75":51,"_76":-5,"_99":531,"_18":532},{"_36":208,"_13":209,"_62":210,"_64":194,"_66":67,"_68":34,"_69":70,"_38":211,"_17":32,"_72":97,"_74":212,"_75":51,"_76":-5,"_99":529,"_18":530},{"_36":201,"_13":202,"_62":203,"_64":194,"_66":67,"_68":34,"_69":70,"_38":204,"_17":32,"_72":97,"_74":205,"_75":51,"_76":-5,"_99":527,"_18":528},{"_36":78,"_13":19,"_62":65,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":145,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":525,"_18":526},{"_36":60,"_13":61,"_62":63,"_64":65,"_66":67,"_68":34,"_69":70,"_38":149,"_17":32,"_72":73,"_74":-5,"_75":51,"_76":-5,"_99":523,"_18":524},{"_36":185,"_13":186,"_62":187,"_64":63,"_66":67,"_68":34,"_69":70,"_38":188,"_17":32,"_72":97,"_74":189,"_75":51,"_76":-5,"_99":521,"_18":522},{"_36":178,"_13":179,"_62":180,"_64":63,"_66":67,"_68":34,"_69":70,"_38":181,"_17":32,"_72":97,"_74":182,"_75":51,"_76":-5,"_99":519,"_18":520},{"_36":171,"_13":172,"_62":173,"_64":63,"_66":67,"_68":34,"_69":70,"_38":174,"_17":32,"_72":97,"_74":175,"_75":51,"_76":-5,"_99":517,"_18":518},{"_36":164,"_13":165,"_62":166,"_64":63,"_66":67,"_68":34,"_69":70,"_38":167,"_17":32,"_72":97,"_74":168,"_75":51,"_76":-5,"_99":515,"_18":516},{"_36":157,"_13":158,"_62":159,"_64":63,"_66":67,"_68":34,"_69":70,"_38":160,"_17":32,"_72":97,"_74":161,"_75":51,"_76":-5,"_99":513,"_18":514},{"_36":138,"_13":139,"_62":140,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":141,"_17":32,"_72":24,"_74":142,"_75":51,"_76":-5,"_99":511,"_18":512},{"_36":131,"_13":132,"_62":133,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":134,"_17":32,"_72":24,"_74":135,"_75":51,"_76":-5,"_99":509,"_18":510},{"_36":124,"_13":125,"_62":126,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":127,"_17":32,"_72":24,"_74":128,"_75":51,"_76":-5,"_99":507,"_18":508},{"_36":117,"_13":118,"_62":119,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":120,"_17":32,"_72":24,"_74":121,"_75":51,"_76":-5,"_99":505,"_18":506},{"_36":110,"_13":111,"_62":112,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":113,"_17":32,"_72":24,"_74":114,"_75":51,"_76":-5,"_99":503,"_18":504},{"_36":102,"_13":103,"_62":104,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":105,"_17":32,"_72":106,"_74":107,"_75":51,"_76":-5,"_99":501,"_18":502},{"_36":93,"_13":94,"_62":95,"_64":-5,"_66":67,"_68":34,"_69":70,"_38":96,"_17":32,"_72":97,"_74":98,"_75":51,"_76":-5,"_99":499,"_18":500},[],[94],[],[103],[],[111],[],[118],[],[125],[],[132],[],[139],[],[19,61,158],[],[19,61,165],[],[19,61,172],[],[19,61,179],[],[19,61,186],[487,488,489,490,491],[19,61],[486],[19],[],[193,202],[],[193,209],[],[193,216],[482,483,484],[193],[],[223,235],[],[223,243],[],[223,250],[],[223,257],[],[223,264],[],[223,271],[475,476,477,478,479,480],[223],[],[278,287],[],[278,294],[],[278,301],[471,472,473],[278],[],[308,321],[],[308,328],[],[308,335],[],[308,342],[],[308,349],[],[308,356],[],[308,363,342],[463],[308,363],[462,464,465,466,467,468,469],[308],["D",1771120376884],["D",1789604441792],{"_11":12,"_422":418},{"_11":12,"_416":-5,"_417":418,"_382":383,"_419":420,"_421":51},{"_17":34,"_13":16,"_36":67,"_42":375,"_45":37,"_46":412,"_413":53,"_414":43,"_415":57,"_58":14,"_54":12},"ctx",{"_406":-5,"_441":613,"_614":615,"_616":617,"_618":51,"_619":51,"_620":51,"_621":70,"_622":-5,"_623":-7,"_624":-7,"_625":32,"_626":51,"_627":51,"_628":-5,"_629":51,"_630":51,"_631":51,"_632":-5},"ENV",{"_584":585,"_586":587,"_588":16,"_589":590,"_591":592,"_593":594,"_595":596,"_597":598,"_599":600,"_601":602,"_603":604,"_605":606,"_607":608,"_609":610,"_611":612},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","c73f21da490985bff19c279eb0e09db24d688e4f","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"]