Sentry Webhook Setup

\n \n

Webhook URL:

${webhookUrl}

\n\n

\n ⚠️ Important: Set the SENTRY_WEBHOOK_SECRET environment variable in Val Town with your webhook client secret before testing.\n

\n\n

Configuration Steps:

Go to Sentry → Settings → Developer Settings → Internal Integrations
Create a new internal integration or edit existing one
Add webhook URL above to \"Webhook URL\" field
Copy the \"Client Secret\" and add it to Val Town as SENTRY_WEBHOOK_SECRET
Enable desired permissions (Issue & Event: Read recommended)
Subscribe to events (Error, Issue recommended)
Save and test by triggering an error

\n\n

The webhook uses AI to analyze error notifications and route them to appropriate engineering channels based on severity and context.

\n\n`;\n\n return new Response(html, {\n headers: { 'Content-Type': 'text/html' }\n });\n }\n\n if (req.method !== 'POST') {\n return new Response('Method not allowed', { status: 405 });\n }\n\n try {\n // Get the raw body and signature\n const body = await req.text();\n const signature = req.headers.get('sentry-hook-signature') || '';\n\n // Verify the webhook signature\n if (!await verifySentrySignature(body, signature, WEBHOOK_CONFIG.sentry.signingSecret)) {\n console.error('Invalid webhook signature');\n return new Response('Unauthorized', { status: 401 });\n }\n\n // Parse the payload\n const payload: SentryWebhookPayload = JSON.parse(body);\n \n // Log the full webhook payload for debugging\n logFormatted(payload, \"Sentry Webhook Received\");\n \n const dataType = payload.data.error ? 'error' : 'issue';\n const dataTitle = payload.data.error?.title || payload.data.issue?.title || 'Unknown';\n console.log(`Processing Sentry ${payload.action} on ${dataType}: ${dataTitle}`);\n\n // Skip certain actions that don't need notifications\n const skipActions = ['resolved', 'ignored'];\n if (skipActions.includes(payload.action)) {\n return new Response('OK - Skipped', { status: 200 });\n }\n\n // Convert to standardized notification format\n const notificationData = convertSentryToNotificationData(payload);\n \n // Process through shared triage system (this will handle its own logging)\n await processNotification(notificationData);\n\n return new Response('OK', { status: 200 });\n\n } catch (error) {\n console.error('Error processing webhook:', error);\n return new Response('Internal Server Error', { status: 500 });\n }\n}",[],[194],"d4be5f02-7f8f-11f0-b611-0224a6c84d84","notification-triage.ts","f2082c72-7c43-11f0-b3d2-0224a6c84d84",["D",1755891617459],23,"/**\n * Shared Notification Triage System\n *\n * This module provides a unified interface for triaging notifications from\n * various sources (Linear, Sentry, billing, etc.) using AI to determine\n * appropriate routing channels.\n */\n\nimport {\n AI_PROMPTS,\n EMAIL_CONFIG,\n NOTIFICATION_CHANNELS,\n type NotificationChannel,\n} from \"./CONFIGURATION.ts\";\nimport { aiService } from \"./ai-service.ts\";\nimport { email } from \"https://esm.town/v/std/email\";\nimport { logFormatted, stringifyForLogging } from \"./stringify-utils.ts\";\nimport { CitationContext } from \"./citation-context.ts\";\nimport { AITimer, globalAITracker, timeAIOperation } from \"./timing-utils.ts\";\nimport { sanitizeNotificationData, logNotificationProcessing, createDataSummary } from \"./data-sanitization.ts\";\n\n/**\n * Standardized notification data structure that all webhook sources\n * should convert their payloads into before triaging\n */\nexport interface NotificationData {\n // Core identification\n source: string; // e.g., 'linear', 'sentry', 'billing'\n id: string; // unique identifier from source system\n type: string; // e.g., 'issue', 'error', 'payment_failed'\n action: string; // e.g., 'created', 'updated', 'resolved'\n\n // Content\n title: string;\n description?: string;\n\n // Metadata\n priority?: number | string; // source-specific priority\n severity?: \"low\" | \"medium\" | \"high\" | \"critical\";\n labels?: string[];\n tags?: string[];\n\n // Context\n team?: string;\n assignee?: {\n name: string;\n email?: string;\n };\n project?: string;\n environment?: string; // e.g., 'production', 'staging'\n\n // Links and references\n url?: string; // primary link to the item\n additionalUrls?: { [key: string]: string }; // other relevant links\n\n // Timestamps\n createdAt?: string;\n updatedAt?: string;\n\n // Raw data for debugging\n rawPayload?: any;\n}\n\n/**\n * Result of AI triage analysis\n */\nexport interface TriageResult {\n selectedChannels: string[];\n reasoning: string;\n summary: string;\n urgency: \"low\" | \"medium\" | \"high\" | \"critical\";\n suggestedActions?: string[];\n}\n\n/**\n * Enhanced triage result with citation context\n */\nexport interface EnhancedTriageResult extends TriageResult {\n citationContext: CitationContext;\n formattedSummary: string; // summary with proper links\n}\n\n/**\n * Channel-specific summary for tailored notifications\n */\nexport interface ChannelSpecificSummary {\n summary: string; // Concise one-line summary tailored for this specific channel\n routingReason: string; // Why this notification was sent to this channel (for logging)\n}\n\n/**\n * Create citation context for a notification to improve AI prompting\n */\nfunction createCitationContext(\n notification: NotificationData,\n): CitationContext {\n const ctx = new CitationContext();\n\n // Generate citation for the main notification\n const mainCitation = ctx.generate(\n notification.source,\n notification.id,\n [notification.title, notification.type, notification.action],\n );\n\n // Add citations for related URLs if they exist\n if (notification.url) {\n ctx.generate(\n \"link\",\n notification.url,\n [`${notification.source} link`, notification.title],\n );\n }\n\n // Add citations for additional URLs\n if (notification.additionalUrls) {\n Object.entries(notification.additionalUrls).forEach(([key, url]) => {\n ctx.generate(\"link\", url, [key, notification.title]);\n });\n }\n\n return ctx;\n}\n\n/**\n * Use AI to determine which channels should receive the notification\n */\nexport async function triageNotification(\n notification: NotificationData,\n): Promise {\n // Log and sanitize notification data first\n const { sanitized: sanitizedNotification, summary: dataSummary } = logNotificationProcessing(notification);\n \n const citationCtx = createCitationContext(notification);\n\n // Prepare channel information\n const channelsInfo = NOTIFICATION_CHANNELS.map((channel) =>\n `${channel.id}: ${channel.name} - ${channel.description} (Priority: ${channel.priority}, Keywords: ${\n channel.keywords.join(\", \")\n })`\n ).join(\"\\n\");\n\n // https://linear.app/developers/webhooks#data-change-events-payload\n if (notification.action === \"update\") {\n logFormatted({\n reason: \"Skipping generic update action\",\n action: notification.action,\n source: notification.source,\n id: notification.id,\n title: notification.title\n }, \"Early Exit Decision\");\n \n return {\n selectedChannels: [],\n reasoning: \"Ignore generic updates - these typically don't require notifications as they're too frequent and low-value\",\n summary: \"Ignored generic update\",\n citationContext: citationCtx,\n formattedSummary: \"Ignored generic update\",\n urgency: \"low\",\n };\n }\n\n // Use sanitized data for AI processing (removes rawPayload and sensitive keys)\n const notificationContext = {\n source: sanitizedNotification.source,\n id: sanitizedNotification.id,\n type: sanitizedNotification.type,\n action: sanitizedNotification.action,\n title: sanitizedNotification.title,\n description: sanitizedNotification.description || \"No description provided\",\n priority: sanitizedNotification.priority || \"Not specified\",\n severity: sanitizedNotification.severity || \"Not specified\",\n labels: sanitizedNotification.labels || [],\n tags: sanitizedNotification.tags || [],\n team: sanitizedNotification.team || \"Unknown\",\n assignee: sanitizedNotification.assignee?.name || \"Unassigned\",\n project: sanitizedNotification.project || \"Unknown\",\n environment: sanitizedNotification.environment || \"Unknown\",\n hasUrl: !!sanitizedNotification.url,\n additionalUrlCount: Object.keys(sanitizedNotification.additionalUrls || {}).length,\n };\n\n // Prepare the context for LLM consumption (replace IDs with citations)\n const contextString = citationCtx.prepareForLLM(\n stringifyForLogging(notificationContext),\n );\n\n const prompt = `${\n AI_PROMPTS.channelSelection.replace(\"{channels}\", channelsInfo)\n }\n\nNotification to triage:\n${contextString}\n\nPlease analyze this notification and respond with a JSON object containing:\n- selectedChannels: array of channel IDs that should receive this notification\n- reasoning: brief explanation of why these channels were selected\n- summary: a concise summary tailored for the selected audience\n- urgency: one of 'low', 'medium', 'high', 'critical'\n- suggestedActions: array of specific actions that recipients should consider (optional)\n\nConsider the notification's source, type, priority, and content when making routing decisions.`;\n\n try {\n // Determine which model to use based on notification complexity\n let useCase: \"fast\" | \"balanced\" | \"advanced\" | \"reasoning\" = \"balanced\";\n\n // Use fast model for simple notifications to avoid timeouts\n if (sanitizedNotification.severity === \"low\" && !sanitizedNotification.description) {\n useCase = \"fast\";\n } // Use fast model for critical issues to ensure quick response\n else if (\n sanitizedNotification.severity === \"critical\" || sanitizedNotification.priority === 1\n ) {\n useCase = \"fast\"; // Changed from 'advanced' to 'fast' to avoid timeouts\n } // Use balanced model for complex scenarios\n else if (\n sanitizedNotification.labels?.length > 3 || sanitizedNotification.tags?.length > 3 ||\n (sanitizedNotification.description && sanitizedNotification.description.length > 500)\n ) {\n useCase = \"balanced\"; // Changed from 'reasoning' to 'balanced'\n }\n\n // Log AI triage input with model selection reasoning\n logFormatted({\n notificationSummary: dataSummary,\n selectedModel: aiService.getModelForUseCase(useCase),\n modelUseCase: useCase,\n modelSelectionReason: {\n severity: sanitizedNotification.severity || 'not specified',\n hasDescription: !!sanitizedNotification.description,\n labelCount: sanitizedNotification.labels?.length || 0,\n descriptionLength: sanitizedNotification.description?.length || 0\n },\n citationEntries: citationCtx.entries(),\n promptLength: prompt.length\n }, \"AI Triage Input & Model Selection\");\n\n // Time the AI operation with enhanced logging\n const { result: triageResult, timing } = await timeAIOperation(\n 'notification_triage',\n async () => {\n return await aiService.completeJSON(\n prompt,\n \"You are a notification triage system. Always respond with valid JSON only. Consider urgency, technical domain, and team context when routing notifications.\",\n {\n useCase,\n maxTokens: 400, // Reduced from 800 to speed up response\n temperature: 0.1,\n },\n );\n },\n aiService.getModelForUseCase(useCase)\n );\n\n // Add timing to global tracker\n globalAITracker.add(timing);\n\n // Create formatted summary with proper links\n const formattedSummary = citationCtx.linkify(\n triageResult.summary,\n \"[key](uri)\",\n );\n\n // Log detailed AI triage result with reasoning explanation\n logFormatted({\n aiDecision: {\n selectedChannels: triageResult.selectedChannels,\n urgency: triageResult.urgency,\n channelNames: NOTIFICATION_CHANNELS\n .filter(ch => triageResult.selectedChannels.includes(ch.id))\n .map(ch => ch.name)\n },\n reasoning: triageResult.reasoning,\n summary: triageResult.summary,\n suggestedActions: triageResult.suggestedActions || [],\n performance: {\n durationMs: timing.durationMs,\n modelUsed: timing.modelId,\n success: timing.success\n },\n dataProcessed: {\n source: sanitizedNotification.source,\n type: sanitizedNotification.type,\n priority: sanitizedNotification.priority,\n severity: sanitizedNotification.severity,\n hasLabels: (sanitizedNotification.labels?.length || 0) > 0,\n descriptionLength: sanitizedNotification.description?.length || 0\n }\n }, \"AI Triage Decision & Reasoning\");\n\n return {\n ...triageResult,\n citationContext: citationCtx,\n formattedSummary,\n };\n } catch (error) {\n console.error(\"Error in AI triage:\", error);\n\n // Log the AI failure with timing information\n const failureTimer = new AITimer('notification_triage_fallback');\n const fallbackTiming = failureTimer.end(false, error.message);\n globalAITracker.add(fallbackTiming);\n\n // Fallback logic based on notification characteristics\n let fallbackChannels = [\"general\"];\n let fallbackUrgency: \"low\" | \"medium\" | \"high\" | \"critical\" = \"medium\";\n\n // Simple fallback routing logic using sanitized notification data\n if (\n sanitizedNotification.severity === \"critical\" || sanitizedNotification.priority === 1 ||\n sanitizedNotification.priority === \"high\"\n ) {\n fallbackChannels = [\"engineering-critical\"];\n fallbackUrgency = \"critical\";\n } else if (\n sanitizedNotification.source === \"sentry\" ||\n sanitizedNotification.environment === \"production\"\n ) {\n fallbackChannels = [\"engineering-general\", \"devops\"];\n fallbackUrgency = \"high\";\n }\n\n // Log the fallback decision with detailed reasoning\n logFormatted({\n fallbackTrigger: {\n error: error.message,\n errorType: error.constructor.name\n },\n fallbackDecision: {\n selectedChannels: fallbackChannels,\n urgency: fallbackUrgency,\n channelNames: NOTIFICATION_CHANNELS\n .filter(ch => fallbackChannels.includes(ch.id))\n .map(ch => ch.name)\n },\n fallbackReasoning: {\n severity: sanitizedNotification.severity,\n priority: sanitizedNotification.priority,\n source: sanitizedNotification.source,\n environment: sanitizedNotification.environment || 'unknown'\n },\n performance: {\n aiFailureTime: fallbackTiming.durationMs,\n totalProcessingAttempted: true\n }\n }, \"AI Triage Fallback Decision\");\n\n return {\n selectedChannels: fallbackChannels,\n reasoning:\n `AI triage failed (${error.message}). Applied fallback routing: ${sanitizedNotification.severity === \"critical\" ? \"Critical severity\" : sanitizedNotification.source === \"sentry\" ? \"Sentry source\" : \"Default routing\"} → ${fallbackChannels.join(\", \")}`,\n summary:\n `${sanitizedNotification.source} ${sanitizedNotification.action}: ${sanitizedNotification.title}`,\n urgency: fallbackUrgency,\n citationContext: citationCtx,\n formattedSummary:\n `${sanitizedNotification.source} ${sanitizedNotification.action}: ${sanitizedNotification.title}`,\n };\n }\n}\n\n/**\n * Generate a channel-specific summary tailored for the target audience\n */\nasync function generateChannelSpecificSummary(\n notification: NotificationData,\n triageResult: EnhancedTriageResult,\n channel: NotificationChannel,\n): Promise {\n try {\n // Create a concise context for the AI to work with\n const notificationContext = {\n title: notification.title,\n source: notification.source,\n type: notification.type,\n action: notification.action,\n severity: notification.severity || \"medium\",\n team: notification.team,\n assignee: notification.assignee?.name,\n // Only include description if it's reasonably short to avoid token bloat\n description: notification.description && notification.description.length <= 200 \n ? notification.description \n : notification.description?.substring(0, 200) + \"...\" || undefined,\n labels: notification.labels?.slice(0, 5), // Limit labels to prevent token bloat\n };\n\n const prompt = `You are writing a notification summary specifically for the \"${channel.name}\" team.\n\nChannel Context:\n- Name: ${channel.name}\n- Description: ${channel.description} \n- Keywords: ${channel.keywords.join(\", \")}\n\nNotification:\n${stringifyForLogging(notificationContext)}\n\nWrite an extremely concise, actionable summary tailored for ${channel.name}. This will be the ONLY content in an email, so it must be complete but brief.\n\nRespond with JSON containing:\n- summary: A single, complete sentence (15-20 words max) that tells ${channel.name} exactly what happened and what they need to know\n- routingReason: Why this notification is relevant to ${channel.name} (for logging only)\n\nExample good summaries:\n- \"Critical payment bug in production affecting checkout flow - needs immediate investigation\" \n- \"New feature request for mobile app dashboard from Product team\"\n- \"Database performance degradation detected on staging environment\"`;\n\n // Use fast model for channel-specific summaries to avoid delays\n const result = await timeAIOperation(\n 'channel_summary',\n async () => {\n return await aiService.completeJSON(\n prompt,\n `You are a notification summarizer. Create ultra-concise, one-line summaries. Always respond with valid JSON only.`,\n {\n useCase: \"fast\", // Use fast model for quick summaries\n maxTokens: 100, // Ultra-concise summaries\n temperature: 0.1,\n },\n );\n },\n aiService.getModelForUseCase(\"fast\")\n );\n\n globalAITracker.add(result.timing);\n\n // Log the channel-specific summary generation\n logFormatted({\n channel: {\n name: channel.name,\n description: channel.description.substring(0, 100) + \"...\",\n },\n summary: result.result,\n performance: {\n durationMs: result.timing.durationMs,\n modelUsed: result.timing.modelId,\n },\n }, `Channel-Specific Summary for ${channel.name}`);\n\n return result.result;\n\n } catch (error) {\n console.error(`Error generating channel-specific summary for ${channel.name}:`, error);\n \n // Fallback to generic summary if AI fails\n return {\n summary: triageResult.formattedSummary,\n routingReason: triageResult.reasoning,\n };\n }\n}\n\n/**\n * Format and send notification to selected channels\n */\nexport async function sendNotifications(\n notification: NotificationData,\n triageResult: EnhancedTriageResult,\n): Promise {\n const selectedChannels = NOTIFICATION_CHANNELS.filter(\n (channel) => triageResult.selectedChannels.includes(channel.id),\n );\n\n // Handle zero channels case\n if (selectedChannels.length === 0) {\n logFormatted({\n notification: {\n source: notification.source,\n id: notification.id,\n title: notification.title,\n },\n decision: \"No channels selected - notification will not be sent\",\n reasoning: triageResult.reasoning,\n }, \"Zero Channel Decision\");\n \n console.log(`📭 No channels selected for ${notification.source} notification: ${notification.title}`);\n return;\n }\n\n logFormatted({\n notification: {\n source: notification.source,\n id: notification.id,\n title: notification.title,\n },\n selectedChannels: selectedChannels.map((c) => c.name),\n reasoning: triageResult.reasoning,\n }, \"Sending Notifications\");\n\n for (const channel of selectedChannels) {\n try {\n // Generate channel-specific summary\n const channelSpecificSummary = await generateChannelSpecificSummary(\n notification, \n triageResult, \n channel\n );\n\n const urgencyPrefix = triageResult.urgency === \"critical\"\n ? \"[CRITICAL] \"\n : triageResult.urgency === \"high\"\n ? \"[HIGH] \"\n : \"\";\n\n const subject =\n `${EMAIL_CONFIG.subjectPrefix} ${urgencyPrefix}${notification.title}`;\n\n // Create formatted links using citation context\n const formattedLinks = triageResult.citationContext.linkify(\n notification.url ? `[#link_1]` : \"No direct link available\",\n '[key](uri \"snippet\")',\n );\n\n const emailBody = `${channelSpecificSummary.summary}\n${formattedLinks}`;\n\n await email({\n to: channel.email,\n subject: subject,\n html: emailBody,\n text: emailBody.replace(/<[^>]*>/g, \"\"), // Strip HTML for text version\n });\n\n console.log(`✅ Notification sent to ${channel.name} (${channel.email})`);\n } catch (error) {\n console.error(\n `❌ Failed to send notification to ${channel.name}:`,\n error,\n );\n }\n }\n}\n\n/**\n * Main triage function that processes any notification data\n */\nexport async function processNotification(\n notification: NotificationData,\n): Promise {\n const processingStartTime = Date.now();\n \n try {\n logFormatted(\n notification,\n `${notification.source.toUpperCase()} Notification Data`,\n );\n\n // Triage the notification (with built-in timing and sanitization)\n const triageResult = await triageNotification(notification);\n \n // Send notifications to selected channels\n await sendNotifications(notification, triageResult);\n\n const totalProcessingTime = Date.now() - processingStartTime;\n\n // Log final processing summary with performance metrics\n logFormatted({\n notification: {\n source: notification.source,\n id: notification.id,\n title: notification.title,\n severity: notification.severity\n },\n result: {\n channelsSelected: triageResult.selectedChannels.length,\n channelNames: NOTIFICATION_CHANNELS\n .filter(ch => triageResult.selectedChannels.includes(ch.id))\n .map(ch => ch.name),\n urgency: triageResult.urgency\n },\n performance: {\n totalProcessingTimeMs: totalProcessingTime,\n aiOperationsCount: globalAITracker.getStats().totalOperations\n }\n }, \"Final Processing Summary\");\n\n // Optionally log AI performance statistics periodically\n if (globalAITracker.getStats().totalOperations % 5 === 0) {\n globalAITracker.logStats();\n }\n\n console.log(\n `✅ Successfully processed ${notification.source} notification: ${notification.title} (${totalProcessingTime}ms total)`,\n );\n } catch (error) {\n const totalProcessingTime = Date.now() - processingStartTime;\n \n logFormatted({\n error: error.message,\n errorType: error.constructor.name,\n notification: {\n source: notification.source,\n id: notification.id,\n title: notification.title\n },\n performance: {\n failureTimeMs: totalProcessingTime,\n aiOperationsAttempted: globalAITracker.getStats().totalOperations\n }\n }, \"Processing Failure Summary\");\n \n console.error(\n `❌ Error processing ${notification.source} notification (${totalProcessingTime}ms):`,\n error,\n );\n throw error;\n }\n}",[],[203],"34097c3e-7c3f-11f0-bc2e-0224a6c84d84","main.tsx","34097ce8-7c3f-11f0-bc2e-0224a6c84d84",["D",1755527134622],"",[],[211],"f752ee5a-7f8c-11f0-8cc1-0224a6c84d84","linear-webhook.ts","ca57e836-7c41-11f0-b28e-0224a6c84d84",99,["D",1755890386995],13,"/**\n * Linear Webhook Handler\n * \n * Receives webhooks from Linear, validates the signature, and processes\n * the notification data for triage.\n */\n\nimport { WEBHOOK_CONFIG } from './CONFIGURATION.ts';\nimport { logFormatted } from './stringify-utils.ts';\nimport { processNotification, type NotificationData } from './notification-triage.ts';\n\ninterface LinearWebhookPayload {\n action: string;\n type: string;\n data: {\n id: string;\n title?: string;\n description?: string;\n state?: {\n name: string;\n type: string;\n };\n priority?: number;\n labels?: Array<{\n name: string;\n }>;\n team?: {\n name: string;\n };\n assignee?: {\n name: string;\n email: string;\n };\n url?: string;\n createdAt?: string;\n updatedAt?: string;\n };\n url: string;\n createdAt: string;\n}\n\n/**\n * Verify Linear webhook signature\n */\nasync function verifyLinearSignature(\n payload: string,\n signature: string,\n secret: string\n): Promise {\n if (!secret) {\n console.warn('Linear webhook secret not configured');\n return false;\n }\n\n try {\n const encoder = new TextEncoder();\n const key = await crypto.subtle.importKey(\n 'raw',\n encoder.encode(secret),\n { name: 'HMAC', hash: 'SHA-256' },\n false,\n ['sign']\n );\n\n const expectedSignature = await crypto.subtle.sign(\n 'HMAC',\n key,\n encoder.encode(payload)\n );\n\n const expectedHex = Array.from(new Uint8Array(expectedSignature))\n .map(b => b.toString(16).padStart(2, '0'))\n .join('');\n\n // Linear sends signature as \"sha256=\"\n const receivedHex = signature.replace('sha256=', '');\n \n return expectedHex === receivedHex;\n } catch (error) {\n console.error('Error verifying signature:', error);\n return false;\n }\n}\n\n/**\n * Convert Linear webhook payload to standardized notification data\n */\nfunction convertLinearToNotificationData(payload: LinearWebhookPayload): NotificationData {\n // Determine severity based on priority and labels\n let severity: 'low' | 'medium' | 'high' | 'critical' = 'medium';\n const labels = payload.data.labels?.map(l => l.name.toLowerCase()) || [];\n \n if (payload.data.priority === 1 || labels.includes('critical') || labels.includes('urgent')) {\n severity = 'critical';\n } else if (payload.data.priority === 2 || labels.includes('high')) {\n severity = 'high';\n } else if (payload.data.priority === 4 || labels.includes('low')) {\n severity = 'low';\n }\n \n return {\n source: 'linear',\n id: payload.data.id,\n type: payload.type.toLowerCase(),\n action: payload.action,\n title: payload.data.title || 'Untitled Linear Item',\n description: payload.data.description,\n priority: payload.data.priority,\n severity,\n labels: payload.data.labels?.map(l => l.name) || [],\n team: payload.data.team?.name,\n assignee: payload.data.assignee ? {\n name: payload.data.assignee.name,\n email: payload.data.assignee.email\n } : undefined,\n url: payload.data.url || payload.url,\n createdAt: payload.data.createdAt || payload.createdAt,\n updatedAt: payload.data.updatedAt,\n rawPayload: payload\n };\n}\n\n/**\n * Process notification asynchronously to avoid webhook timeouts\n */\nasync function processNotificationAsync(notificationData: NotificationData): Promise {\n try {\n // Process through shared triage system (this will handle its own logging)\n await processNotification(notificationData);\n console.log(`✅ Async processing completed for Linear notification: ${notificationData.title}`);\n } catch (error) {\n console.error(`❌ Async processing failed for Linear notification: ${notificationData.title}`, error);\n }\n}\n\n/**\n * Main webhook handler\n */\nexport default async function(req: Request): Promise {\n // Handle GET requests with configuration instructions\n if (req.method === 'GET') {\n const url = new URL(req.url);\n const webhookUrl = url.origin + url.pathname;\n \n const html = `\n\n\n\n Linear Webhook Configuration\n \n\n\n

🙋 Linear Webhook Setup

\n \n

Webhook URL:

${webhookUrl}

\n\n

\n ⚠️ Important: Set the LINEAR_WEBHOOK_SECRET environment variable in Val Town with your webhook signing secret before testing.\n

\n\n

Configuration Steps:

Go to Linear → Settings → API → Webhooks
Click \"Create webhook\"
Paste the URL above
Copy the signing secret and add it to Val Town as LINEAR_WEBHOOK_SECRET
Select your desired events (Issues, Comments recommended)
Save and test by creating/updating an issue

\n\n

The webhook uses AI to analyze notifications and route them to appropriate email channels based on content and priority.

\n\n`;\n\n return new Response(html, {\n headers: { 'Content-Type': 'text/html' }\n });\n }\n\n if (req.method !== 'POST') {\n return new Response('Method not allowed', { status: 405 });\n }\n\n try {\n // Get the raw body and signature\n const body = await req.text();\n const signature = req.headers.get('linear-signature') || '';\n\n // Verify the webhook signature\n if (!await verifyLinearSignature(body, signature, WEBHOOK_CONFIG.linear.signingSecret)) {\n console.error('Invalid webhook signature');\n return new Response('Unauthorized', { status: 401 });\n }\n\n // Parse the payload\n const payload: LinearWebhookPayload = JSON.parse(body);\n \n // Log the full webhook payload for debugging\n logFormatted(payload, \"Linear Webhook Received\");\n \n console.log(`Processing Linear ${payload.action} on ${payload.type}: ${payload.data.title || 'Untitled'}`);\n\n // Skip certain actions that don't need notifications\n const skipActions = ['view', 'read'];\n if (skipActions.includes(payload.action)) {\n return new Response('OK - Skipped', { status: 200 });\n }\n\n // Convert to standardized notification format\n const notificationData = convertLinearToNotificationData(payload);\n \n // IMPORTANT: Process asynchronously to avoid webhook timeout\n // Don't await this - let it run in the background\n processNotificationAsync(notificationData);\n\n // Return immediately to Linear to prevent timeout\n return new Response('OK - Processing', { status: 200 });\n\n } catch (error) {\n console.error('Error processing webhook:', error);\n return new Response('Internal Server Error', { status: 500 });\n }\n}",[],[218],"b981da96-7f8c-11f0-a464-0224a6c84d84","deno.json","b981db4a-7f8c-11f0-a464-0224a6c84d84",97,["D",1755890283210],"{\n \"$schema\": \"https://raw.githubusercontent.com/denoland/deno/348900b8b79f4a434cab4c74b3bc8d4d2fa8ee74/cli/schemas/config-file.v1.json\",\n \"lock\": false,\n \"compilerOptions\": {\n \"noImplicitAny\": false,\n \"strict\": false,\n \"types\": [\n \"https://www.val.town/types/valtown.d.ts\"\n ],\n \"lib\": [\n \"dom\",\n \"dom.iterable\",\n \"dom.asynciterable\",\n \"deno.ns\",\n \"deno.unstable\"\n ]\n },\n \"lint\": {\n \"files\": {\n \"include\": [\n \"deno:/https/esm.town/**/*\"\n ]\n },\n \"rules\": {\n \"exclude\": [\n \"no-explicit-any\"\n ]\n }\n },\n \"node_modules_dir\": false,\n \"experimental\": {\n \"unstable-node-globals\": true,\n \"unstable-temporal\": true,\n \"unstable-worker-options\": true,\n \"unstable-sloppy-imports\": true\n }\n}",[],[227],"a25a103a-7f8d-11f0-86a2-0224a6c84d84","data-sanitization.ts","a25a113e-7f8d-11f0-86a2-0224a6c84d84",102,["D",1755890673936],"/**\n * Data sanitization utilities\n * \n * This module provides utilities for sanitizing notification data before\n * sending to AI, removing raw/sensitive data while preserving useful context.\n */\n\nimport { NotificationData } from './notification-triage.ts';\nimport { logFormatted } from './stringify-utils.ts';\n\n/**\n * Keys to remove from notification data before AI processing\n * These typically contain raw payload data, sensitive info, or verbose data\n */\nconst SENSITIVE_KEYS = [\n 'rawPayload',\n 'raw',\n 'payload',\n 'webhookData',\n 'originalData',\n 'fullPayload',\n 'credentials',\n 'tokens',\n 'secrets',\n 'apiKeys',\n 'passwords'\n];\n\n/**\n * Keys to remove from nested objects (recursive sanitization)\n */\nconst NESTED_SENSITIVE_KEYS = [\n 'signature',\n 'secret',\n 'token',\n 'key',\n 'credential',\n 'auth',\n 'password',\n 'private'\n];\n\n/**\n * Sanitized notification data for AI processing\n */\nexport interface SanitizedNotificationData extends Omit {\n // Explicitly exclude rawPayload and other sensitive data\n sanitizedAt: string;\n originalKeys?: string[]; // Track what keys were present before sanitization\n}\n\n/**\n * Recursively remove sensitive keys from an object\n */\nfunction sanitizeObject(obj: any, depth = 0, maxDepth = 5): any {\n if (depth > maxDepth) {\n return '[Max depth reached]';\n }\n\n if (obj === null || typeof obj !== 'object') {\n return obj;\n }\n\n if (Array.isArray(obj)) {\n return obj.map(item => sanitizeObject(item, depth + 1, maxDepth));\n }\n\n const sanitized: any = {};\n \n for (const [key, value] of Object.entries(obj)) {\n const lowerKey = key.toLowerCase();\n \n // Skip sensitive keys\n if (NESTED_SENSITIVE_KEYS.some(sensitiveKey => lowerKey.includes(sensitiveKey))) {\n continue;\n }\n\n // Recursively sanitize nested objects\n sanitized[key] = sanitizeObject(value, depth + 1, maxDepth);\n }\n\n return sanitized;\n}\n\n/**\n * Sanitize notification data for AI processing\n * Removes raw payloads, sensitive keys, and limits data size\n */\nexport function sanitizeNotificationData(notification: NotificationData): SanitizedNotificationData {\n const startTime = Date.now();\n \n // Track original keys for debugging\n const originalKeys = Object.keys(notification);\n \n // Create a copy and remove sensitive top-level keys\n const sanitized: any = {};\n \n for (const [key, value] of Object.entries(notification)) {\n const lowerKey = key.toLowerCase();\n \n // Skip sensitive keys entirely\n if (SENSITIVE_KEYS.some(sensitiveKey => lowerKey.includes(sensitiveKey))) {\n continue;\n }\n \n // For nested objects, apply recursive sanitization\n if (typeof value === 'object' && value !== null) {\n sanitized[key] = sanitizeObject(value);\n } else {\n sanitized[key] = value;\n }\n }\n\n // Add sanitization metadata\n sanitized.sanitizedAt = new Date().toISOString();\n sanitized.originalKeys = originalKeys;\n\n const sanitizationTime = Date.now() - startTime;\n\n // Log sanitization details\n logFormatted({\n source: notification.source,\n id: notification.id,\n originalKeyCount: originalKeys.length,\n sanitizedKeyCount: Object.keys(sanitized).length,\n removedKeys: originalKeys.filter(key => !(key in sanitized)),\n sanitizationTimeMs: sanitizationTime\n }, 'Data Sanitization Summary');\n\n return sanitized as SanitizedNotificationData;\n}\n\n/**\n * Create a summary of notification data for enhanced logging\n * This provides a concise view of what data is being processed\n */\nexport function createDataSummary(notification: NotificationData): {\n core: {\n source: string;\n type: string;\n action: string;\n title: string;\n };\n metadata: {\n hasPriority: boolean;\n hasSeverity: boolean;\n hasDescription: boolean;\n hasLabels: boolean;\n hasTeam: boolean;\n hasAssignee: boolean;\n hasUrl: boolean;\n };\n sizes: {\n titleLength: number;\n descriptionLength: number;\n labelCount: number;\n totalKeys: number;\n };\n timestamps: {\n hasCreatedAt: boolean;\n hasUpdatedAt: boolean;\n };\n} {\n return {\n core: {\n source: notification.source,\n type: notification.type,\n action: notification.action,\n title: notification.title\n },\n metadata: {\n hasPriority: notification.priority !== undefined,\n hasSeverity: notification.severity !== undefined,\n hasDescription: !!notification.description,\n hasLabels: !!(notification.labels && notification.labels.length > 0),\n hasTeam: !!notification.team,\n hasAssignee: !!notification.assignee,\n hasUrl: !!notification.url\n },\n sizes: {\n titleLength: notification.title.length,\n descriptionLength: notification.description?.length || 0,\n labelCount: notification.labels?.length || 0,\n totalKeys: Object.keys(notification).length\n },\n timestamps: {\n hasCreatedAt: !!notification.createdAt,\n hasUpdatedAt: !!notification.updatedAt\n }\n };\n}\n\n/**\n * Enhanced logging for notifications that includes both summary and sanitization\n */\nexport function logNotificationProcessing(notification: NotificationData): {\n sanitized: SanitizedNotificationData;\n summary: ReturnType;\n} {\n const summary = createDataSummary(notification);\n const sanitized = sanitizeNotificationData(notification);\n\n logFormatted(summary, 'Notification Data Summary');\n\n return { sanitized, summary };\n}\n",[],[235],"3c58e696-7c57-11f0-8fef-0224a6c84d84","citation-context.ts","d564bf04-7c43-11f0-b301-0224a6c84d84",82,["D",1755537456482],"/**\n * Citation Context for AI Prompting\n * \n * This module provides utilities for generating and linking citation keys\n * in AI prompts, making it easier to reference specific events, resources,\n * and data points in a structured way.\n */\n\nconst warn = console.warn.bind(console, \"[citation context]\");\n\nconst ID_REGEX =\n /^\\b[0-9a-fA-F]{16,32}\\b|\\b[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}\\b$/;\n\n/** Internal structure used by `CitationContext`. */\ntype CitationRecord = {\n uri: string; // stable URI pointing to resource in backend (could be uuid scheme)\n contentSnippet: string; // optional preview/content\n uniqueId?: string; // optional UUID/GUID/etc for replacing keys into another text map\n};\n\n/**\n * CitationContext is a utility class for generating and linking citation keys.\n * It helps track and format references to specific events, runs, or other resources.\n *\n * @example\n * ```ts\n * const citationCtx = new CitationContext();\n * const key = citationCtx.generate('event', '123', ['Event 1', 'Event 2']);\n * const linkedText = citationCtx.linkify('[#event_1, #event_2]', '[key](uri)');\n * // linkedText: [event_1](ch-event://123) | [event_2](ch-event://123)\n * ```\n */\nexport class CitationContext {\n private counters = new Map();\n readonly map = new Map();\n\n /**\n * Generate a citation key and register its mapping.\n * Example output: [#event_3]\n * @param prefix Logical prefix e.g. 'event', 'run'\n * @param uri Stable URI/identifier (can be UUID) for backend lookup\n * @param contentSnippet Optional short snippet for output files/debugging\n */\n generate(prefix: string, id: string, contentSnippet: string[]): string {\n const idx = (this.counters.get(prefix) ?? 0) + 1;\n this.counters.set(prefix, idx);\n const key = `${prefix}_${idx}`;\n this.map.set(key, {\n uri: /^[\\w-]+:\\/\\//.test(id) ? id : `ch-${prefix}://${id}`,\n contentSnippet: contentSnippet.join(' | '),\n // If the id is likely a completely unique ID, use it as the uniqueId\n uniqueId: id.match(ID_REGEX)?.[0],\n });\n return `[#${key}]`;\n }\n\n /**\n * Prepare text for LLM consumption by replacing unique IDs with citation keys.\n *\n * @param text - The text to process\n * @returns The text with unique IDs replaced by citation keys\n */\n prepareForLLM(text: string): string {\n for (const [key, citation] of this.map.entries()) {\n if (citation.uniqueId) {\n text = text.replace(\n new RegExp(`\\\\b${citation.uniqueId}\\\\b`, 'g'),\n `[${key}]`\n );\n }\n }\n return text;\n }\n\n /**\n * Convert inline citation keys inside text to markdown links.\n * Transforms patterns like [#event_1, #event_2] to markdown links.\n *\n * @param text - The text containing citation keys to linkify\n * @param format - The output format for the links\n * @returns The text with citation keys converted to markdown links\n */\n linkify(text: string, format: '[key](uri)' | '[key](uri \"snippet\")'): string {\n return text.replace(\n /(:?\\[#([a-zA-Z]+_\\d+)(?:(?:,\\s*|\\s*[-––|]\\s*)#([a-zA-Z]+_\\d+))*\\])+/g,\n (match) => {\n // Extract all citation keys from the match\n const keys: string[] = match.match(/[a-zA-Z]+_\\d+/g) || [];\n\n // Transform each key into a markdown link\n return keys\n .map((k) => {\n const citation = this.map.get(k);\n if (!citation) {\n const matchedFormat = /^(\\w+)_(\\d+)$/.exec(k);\n let keys = [...this.map.keys()];\n let emptyMessage = 'none';\n if (matchedFormat) {\n const [, prefix] = matchedFormat;\n keys = keys.filter((a) => a.startsWith(prefix));\n emptyMessage = `no ${prefix} citations`;\n }\n warn(\n new Error(\n `Citation (${k}) did not exist from [${keys.join(', ') || emptyMessage}]`\n ),\n 'Citation key not found in map'\n );\n return `[#${k}]`; // Handle case where key doesn't exist in map\n }\n\n switch (format) {\n case '[key](uri)':\n return `[${k}](${citation.uri})`;\n case '[key](uri \"snippet\")':\n default:\n return `[${k}](${citation.uri} \"${citation.contentSnippet?.replace(/[\"\\n]+/g, '')}\")`;\n }\n })\n .join(', ');\n }\n );\n }\n\n /**\n * Get all citation entries as an array of key-value pairs.\n *\n * @returns Array of [key, CitationRecord] tuples\n */\n entries(): [string, CitationRecord][] {\n return Array.from(this.map.entries());\n }\n}",[],[243],"da4f6062-7f8d-11f0-a4d6-0224a6c84d84","ai-service.ts","acd4aaca-7c45-11f0-bb97-0224a6c84d84",109,["D",1755890767817],8,"/**\n * AI Service using OpenRouter\n * \n * This module provides a unified interface for AI operations using OpenRouter\n * as the provider, with configurable models and fallback support.\n */\n\nimport { AI_CONFIG, AI_MODELS, getModelConfig, getModelByUseCase } from './CONFIGURATION.ts';\nimport { logFormatted } from './stringify-utils.ts';\n\nexport interface ChatMessage {\n role: 'system' | 'user' | 'assistant';\n content: string;\n}\n\nexport interface ChatCompletionRequest {\n messages: ChatMessage[];\n model?: string;\n temperature?: number;\n maxTokens?: number;\n topP?: number;\n useCase?: 'fast' | 'balanced' | 'advanced' | 'reasoning';\n provider?: 'groq' | 'openai' | 'anthropic' | 'auto';\n}\n\nexport interface ChatCompletionResponse {\n id: string;\n model: string;\n choices: Array<{\n message: {\n role: string;\n content: string;\n };\n finishReason: string;\n }>;\n usage: {\n promptTokens: number;\n completionTokens: number;\n totalTokens: number;\n };\n}\n\n/**\n * OpenRouter AI Service\n */\nexport class OpenRouterService {\n private apiKey: string;\n private baseUrl: string;\n private siteName: string;\n private siteUrl: string;\n\n constructor() {\n this.apiKey = AI_CONFIG.openRouter.apiKey;\n this.baseUrl = AI_CONFIG.openRouter.baseUrl;\n this.siteName = AI_CONFIG.openRouter.siteName;\n this.siteUrl = AI_CONFIG.openRouter.siteUrl;\n\n if (!this.apiKey) {\n throw new Error('OPENROUTER_API_KEY environment variable is required');\n }\n }\n\n /**\n * Make a chat completion request to OpenRouter\n */\n async chatCompletion(request: ChatCompletionRequest): Promise {\n // Determine which model to use\n let modelId = request.model;\n if (!modelId && request.useCase) {\n modelId = getModelByUseCase(request.useCase);\n }\n if (!modelId) {\n modelId = AI_CONFIG.modelSelection.default;\n }\n\n const modelConfig = getModelConfig(modelId);\n if (!modelConfig) {\n throw new Error(`Unknown model: ${modelId}`);\n }\n\n // Prepare request parameters\n const params = {\n model: modelId,\n messages: request.messages,\n temperature: request.temperature ?? AI_CONFIG.defaultParams.temperature,\n max_tokens: Math.min(\n request.maxTokens ?? AI_CONFIG.defaultParams.maxTokens,\n modelConfig.maxTokens\n ),\n top_p: request.topP ?? AI_CONFIG.defaultParams.topP,\n // Set provider - use requested provider or default to groq\n provider: request.provider ? \n AI_CONFIG.providers[request.provider] : \n AI_CONFIG.defaultProvider\n };\n\n const headers = {\n 'Authorization': `Bearer ${this.apiKey}`,\n 'Content-Type': 'application/json',\n 'HTTP-Referer': this.siteUrl,\n 'X-Title': this.siteName,\n };\n\n logFormatted({\n model: modelId,\n modelConfig: {\n name: modelConfig.name,\n provider: modelConfig.provider,\n useCase: modelConfig.useCase\n },\n requestParams: {\n temperature: params.temperature,\n maxTokens: params.max_tokens,\n messageCount: params.messages.length,\n requestedProvider: request.provider || 'groq (default)',\n providerConfig: params.provider\n }\n }, 'OpenRouter Request');\n\n try {\n // Add timeout to prevent webhook timeouts\n const controller = new AbortController();\n const timeoutId = setTimeout(() => controller.abort(), 25000); // 25 second timeout\n\n const response = await fetch(`${this.baseUrl}/chat/completions`, {\n method: 'POST',\n headers,\n body: JSON.stringify(params),\n signal: controller.signal,\n });\n\n clearTimeout(timeoutId);\n\n if (!response.ok) {\n const errorText = await response.text();\n throw new Error(`OpenRouter API error (${response.status}): ${errorText}`);\n }\n\n const result = await response.json() as ChatCompletionResponse;\n \n // Log usage information with timing\n const requestDuration = Date.now() - Date.now(); // This will be overridden by timing utils, but keeping for backward compatibility\n if (result.usage) {\n const inputCost = (result.usage.promptTokens / 1000) * modelConfig.costPer1kTokens.input;\n const outputCost = (result.usage.completionTokens / 1000) * modelConfig.costPer1kTokens.output;\n const totalCost = inputCost + outputCost;\n\n logFormatted({\n model: modelId,\n usage: result.usage,\n estimatedCost: {\n input: `$${inputCost.toFixed(6)}`,\n output: `$${outputCost.toFixed(6)}`,\n total: `$${totalCost.toFixed(6)}`\n },\n performance: {\n tokensPerSecond: result.usage.totalTokens / Math.max((requestDuration / 1000), 0.1),\n costPerSecond: totalCost / Math.max((requestDuration / 1000), 0.1)\n }\n }, 'OpenRouter Usage & Performance');\n }\n\n return result;\n\n } catch (error) {\n console.error(`Error with model ${modelId}:`, error);\n \n // Try fallback models if enabled\n if (AI_CONFIG.fallback.enabled && !request.model) {\n for (const fallbackModel of AI_CONFIG.fallback.models) {\n if (fallbackModel === modelId) continue; // Skip the model that just failed\n \n console.log(`Trying fallback model: ${fallbackModel}`);\n try {\n return await this.chatCompletion({\n ...request,\n model: fallbackModel\n });\n } catch (fallbackError) {\n console.error(`Fallback model ${fallbackModel} also failed:`, fallbackError);\n }\n }\n }\n \n throw error;\n }\n }\n\n /**\n * Simple text completion helper\n */\n async complete(\n prompt: string, \n systemMessage?: string,\n options?: Partial\n ): Promise {\n const messages: ChatMessage[] = [];\n \n if (systemMessage) {\n messages.push({ role: 'system', content: systemMessage });\n }\n \n messages.push({ role: 'user', content: prompt });\n\n const response = await this.chatCompletion({\n messages,\n ...options\n });\n\n return response.choices[0]?.message?.content || '';\n }\n\n /**\n * JSON completion helper - ensures valid JSON response\n */\n async completeJSON(\n prompt: string,\n systemMessage?: string,\n options?: Partial\n ): Promise {\n const fullSystemMessage = systemMessage \n ? `${systemMessage}\\n\\nIMPORTANT: Always respond with valid JSON only. Do not include any text before or after the JSON object. Do not wrap the JSON in markdown code blocks.`\n : 'Always respond with valid JSON only. Do not include any text before or after the JSON object. Do not wrap the JSON in markdown code blocks.';\n\n const response = await this.complete(prompt, fullSystemMessage, options);\n \n // Clean the response to handle markdown code blocks\n let cleanedResponse = response.trim();\n \n // Remove markdown code block formatting if present\n if (cleanedResponse.startsWith('```json')) {\n cleanedResponse = cleanedResponse.replace(/^```json\\s*/, '').replace(/\\s*```$/, '');\n } else if (cleanedResponse.startsWith('```')) {\n cleanedResponse = cleanedResponse.replace(/^```\\s*/, '').replace(/\\s*```$/, '');\n }\n \n // Remove any leading/trailing whitespace again\n cleanedResponse = cleanedResponse.trim();\n \n try {\n return JSON.parse(cleanedResponse) as T;\n } catch (error) {\n console.error('Failed to parse JSON response:', response);\n console.error('Cleaned response:', cleanedResponse);\n throw new Error(`Invalid JSON response from AI: ${error.message}`);\n }\n }\n\n /**\n * Get available models\n */\n getAvailableModels(): typeof AI_MODELS {\n return AI_MODELS;\n }\n\n /**\n * Get model by use case\n */\n getModelForUseCase(useCase: 'fast' | 'balanced' | 'advanced' | 'reasoning'): string {\n return getModelByUseCase(useCase);\n }\n}\n\n// Export singleton instance\nexport const aiService = new OpenRouterService();",[],[251],"7c9da25e-7c49-11f0-a058-0224a6c84d84","ai-model-test.ts",81,["D",1755531551351],4,"/**\n * AI Model Testing Utility\n * \n * This HTTP endpoint helps test different AI models and their performance\n * for notification triage tasks.\n * \n * Usage:\n * - GET / - Show available models and run basic tests\n * - GET /?test=models - Test all models\n * - GET /?test=usecases - Test models by use case\n * - GET /?test=triage - Test notification triage scenario\n * - GET /?model=model-id&prompt=your-prompt - Test specific model\n */\n\nimport { aiService } from './ai-service.ts';\nimport { AI_MODELS, getModelByUseCase } from './CONFIGURATION.ts';\nimport { logFormatted } from './stringify-utils.ts';\n\n/**\n * Test a simple completion with a specific model\n */\nexport async function testModel(\n modelId: string, \n prompt: string = \"Hello! Please respond with a simple greeting.\",\n provider?: 'groq' | 'openai' | 'anthropic' | 'auto'\n): Promise<{ success: boolean; response?: string; error?: string; duration?: number; provider?: string }> {\n try {\n const start = Date.now();\n const response = await aiService.complete(prompt, undefined, { model: modelId, provider });\n const duration = Date.now() - start;\n \n return {\n success: true,\n response,\n duration,\n provider: provider || 'groq (default)'\n };\n \n } catch (error) {\n return {\n success: false,\n error: error.message,\n provider: provider || 'groq (default)'\n };\n }\n}\n\n/**\n * Test all available models with a simple prompt\n */\nexport async function testAllModels(): Promise {\n const testPrompt = \"Respond with exactly this JSON: {\\\"status\\\": \\\"ok\\\", \\\"model\\\": \\\"your-model-name\\\"}\";\n const results = [];\n \n for (const model of AI_MODELS) {\n const result = await testModel(model.id, testPrompt);\n results.push({\n modelId: model.id,\n modelName: model.name,\n provider: model.provider,\n useCase: model.useCase,\n ...result\n });\n \n // Small delay between requests to be respectful\n await new Promise(resolve => setTimeout(resolve, 1000));\n }\n \n return results;\n}\n\n/**\n * Test models by use case\n */\nexport async function testByUseCase(): Promise {\n const useCases: Array<'fast' | 'balanced' | 'advanced' | 'reasoning'> = ['fast', 'balanced', 'advanced', 'reasoning'];\n const results = [];\n \n for (const useCase of useCases) {\n const modelId = getModelByUseCase(useCase);\n const prompt = `This is a ${useCase} test. Respond with JSON: {\"useCase\": \"${useCase}\", \"understood\": true}`;\n \n const result = await testModel(modelId, prompt);\n results.push({\n useCase,\n modelId,\n ...result\n });\n \n await new Promise(resolve => setTimeout(resolve, 1000));\n }\n \n return results;\n}\n\n/**\n * Test notification triage with different models\n */\nexport async function testTriageScenario(): Promise {\n const mockNotification = {\n source: 'sentry',\n type: 'error',\n action: 'created',\n title: 'Database connection timeout in production',\n description: 'Multiple users experiencing 500 errors due to database timeouts',\n severity: 'high' as const,\n environment: 'production',\n labels: ['database', 'timeout', 'production']\n };\n \n const prompt = `Analyze this notification and determine urgency (low/medium/high/critical):\n${JSON.stringify(mockNotification, null, 2)}\n\nRespond with JSON: {\"urgency\": \"high\", \"reasoning\": \"your reasoning here\"}`;\n \n // Test with different model types\n const testCases = [\n { useCase: 'fast' as const, description: 'Fast model for simple triage' },\n { useCase: 'balanced' as const, description: 'Balanced model for standard triage' },\n { useCase: 'advanced' as const, description: 'Advanced model for complex analysis' }\n ];\n \n const results = [];\n \n for (const testCase of testCases) {\n const modelId = getModelByUseCase(testCase.useCase);\n \n try {\n const start = Date.now();\n const result = await aiService.completeJSON(prompt, undefined, { useCase: testCase.useCase });\n const duration = Date.now() - start;\n \n results.push({\n useCase: testCase.useCase,\n description: testCase.description,\n modelId,\n success: true,\n result,\n duration\n });\n \n } catch (error) {\n results.push({\n useCase: testCase.useCase,\n description: testCase.description,\n modelId,\n success: false,\n error: error.message\n });\n }\n \n await new Promise(resolve => setTimeout(resolve, 1000));\n }\n \n return results;\n}\n\n/**\n * Show model information and pricing\n */\nexport function getModelInfo(): any {\n const useCases: Array<'fast' | 'balanced' | 'advanced' | 'reasoning'> = ['fast', 'balanced', 'advanced', 'reasoning'];\n \n return {\n availableModels: AI_MODELS.map(model => ({\n id: model.id,\n name: model.name,\n provider: model.provider,\n useCase: model.useCase,\n maxTokens: model.maxTokens,\n costPer1kTokens: model.costPer1kTokens,\n description: model.description\n })),\n modelSelection: Object.fromEntries(\n useCases.map(useCase => {\n const modelId = getModelByUseCase(useCase);\n const model = AI_MODELS.find(m => m.id === modelId);\n return [useCase, { modelId, modelName: model?.name }];\n })\n )\n };\n}\n\n// HTTP endpoint handler\nexport default async function(req: Request): Promise {\n const url = new URL(req.url);\n const method = req.method;\n\n try {\n let result: any;\n let title: string;\n\n if (method === 'POST') {\n // Handle form submissions\n const formData = await req.formData();\n const testType = formData.get('testType') as string;\n const modelId = formData.get('modelId') as string;\n const prompt = formData.get('prompt') as string;\n\n if (testType === 'custom' && modelId && prompt) {\n title = `Testing ${modelId}`;\n result = await testModel(modelId, prompt);\n } else if (testType === 'models') {\n title = 'Testing All Models';\n result = await testAllModels();\n } else if (testType === 'usecases') {\n title = 'Testing Models by Use Case';\n result = await testByUseCase();\n } else if (testType === 'triage') {\n title = 'Testing Notification Triage Scenario';\n result = await testTriageScenario();\n } else {\n throw new Error('Invalid form submission');\n }\n } else {\n // Handle GET requests (legacy support)\n const testType = url.searchParams.get('test');\n const modelId = url.searchParams.get('model');\n const prompt = url.searchParams.get('prompt');\n\n if (modelId && prompt) {\n title = `Testing ${modelId}`;\n result = await testModel(modelId, prompt);\n } else if (testType === 'models') {\n title = 'Testing All Models';\n result = await testAllModels();\n } else if (testType === 'usecases') {\n title = 'Testing Models by Use Case';\n result = await testByUseCase();\n } else if (testType === 'triage') {\n title = 'Testing Notification Triage Scenario';\n result = await testTriageScenario();\n } else {\n // Default: show model info and basic test\n title = 'AI Model Information & Testing';\n const modelInfo = getModelInfo();\n const basicTest = await testModel('anthropic/claude-3.5-sonnet', 'Hello! Please respond with \"AI system is working correctly.\"');\n \n result = {\n modelInfo,\n basicTest\n };\n }\n }\n\n // Return HTML response for better readability\n const html = `\n\n\n\n AI Model Testing - ${title}\n \n \n\n\n

🤖 ${title}

\n \n

\n Quick Legacy Links (GET requests):
\n Model Info & Basic Test\n Test by Use Case\n Test Triage Scenario\n Test All Models\n

\n\n

🧪 Interactive Testing Forms

\n \n

🚀 Custom Model Test

\n \n

\n\n

⚡ Test All Models

\n \n

\n\n

🎯 Test by Use Case

\n \n

\n\n

🔍 Test Triage Scenario

\n \n

\n\n

📊 Available Models

\n ${AI_MODELS.map(model => `\n

${model.name}

by ${model.provider}

${model.useCase}

\n 💰 Input: ${model.costPer1kTokens.input}/1K tokens
\n 💰 Output: ${model.costPer1kTokens.output}/1K tokens
\n 📏 Max tokens: ${model.maxTokens.toLocaleString()}\n

${model.description}

\n
\n \n

\n `).join('')}\n

\n\n

📈 Current Model Selection

\n ${Object.entries({\n 'Fast': getModelByUseCase('fast'),\n 'Balanced': getModelByUseCase('balanced'), \n 'Advanced': getModelByUseCase('advanced'),\n 'Reasoning': getModelByUseCase('reasoning')\n }).map(([useCase, modelId]) => {\n const model = AI_MODELS.find(m => m.id === modelId);\n return model ? `\n

${useCase} → ${model.name}

by ${model.provider}

\n 💰 ${model.costPer1kTokens.input}/${model.costPer1kTokens.output} per 1K tokens
\n 📏 ${model.maxTokens.toLocaleString()} max tokens\n

\n \n

\n ` : '';\n }).join('')}\n

\n\n

📋 Test Results

${JSON.stringify(result, null, 2)}

\n \n

💡 Usage Tips

Quick Test Prompts:

JSON Test: \"Respond with exactly this JSON: {\\\"status\\\": \\\"ok\\\", \\\"model\\\": \\\"your-model-name\\\"}\"
Triage Test: \"Analyze this notification: Critical database timeout in production affecting 500 users. Respond with JSON containing urgency and reasoning.\"
Capability Test: \"What are your key strengths and ideal use cases? Respond in 2-3 sentences.\"
Speed Test: \"Hello! Please respond with just 'OK' to test response time.\"

\n \n

Model Selection Guide:

Fast (GPT-OSS 20B): Simple notifications, quick responses, cost-effective
Balanced (Claude 3.5 Sonnet): Standard triage, good quality/cost balance
Advanced (GPT-4o): Critical issues requiring sophisticated analysis
Reasoning (Kimi K2): Complex scenarios, tool use, coding tasks

\n\n`;\n\n return new Response(html, {\n headers: { 'Content-Type': 'text/html' }\n });\n\n } catch (error) {\n return new Response(JSON.stringify({\n error: error.message,\n stack: error.stack\n }, null, 2), {\n status: 500,\n headers: { 'Content-Type': 'application/json' }\n });\n }\n}",[],[260],"54f9a136-7c84-11f0-8db0-0224a6c84d84","README.md","54f62abe-7c3f-11f0-9d34-0224a6c84d84",92,["D",1755556825138],"# Notification Triage Webhook System\n\nAn intelligent notification routing system that receives webhooks from various sources (Linear, Sentry, etc.) and uses AI to determine the most appropriate channels for delivery. The system analyzes notification content and routes messages to predefined email channels based on context, priority, and team relevance.\n\n## Core Concept\n\nInstead of broadcasting all notifications to everyone, this system:\n1. **Receives** webhooks from external services\n2. **Analyzes** the content using AI (via OpenRouter with multiple model options)\n3. **Routes** notifications to appropriate channels based on:\n - Content analysis and keywords\n - Priority/severity levels\n - Team assignments\n - Predefined channel descriptions\n4. **Delivers** formatted, actionable messages via email\n\nEach notification includes:\n- Clear summary of what happened\n- Relevant links in proper format\n- Context tailored to the target audience\n- Action items when applicable\n\n## AI-First Approach with OpenRouter\n\nThis system uses **OpenRouter** as the AI provider, giving access to multiple AI models with intelligent model selection based on notification complexity and cost optimization.\n\n### Multi-Model AI Strategy\n\nThe system automatically selects the most appropriate AI model based on the notification characteristics:\n\n- **Fast Model** (`openai/gpt-oss-20b`): Simple notifications with low severity and minimal description - ultra-cheap at $0.10/$0.50 per million tokens\n- **Balanced Model** (`anthropic/claude-3.5-sonnet`): Standard triage operations (default)\n- **Advanced Model** (`openai/gpt-4o`): Critical issues requiring sophisticated analysis\n- **Reasoning Model** (`moonshotai/kimi-k2`): Complex scenarios with multiple factors, excellent for tool use and coding\n\n### Available Models\n\n| Model | Provider | Use Case | Cost (per 1M tokens) | Max Tokens | Context |\n|-------|----------|----------|---------------------|------------|---------|\n| GPT-OSS 20B | OpenAI | Fast | $0.10/$0.50 | 131K | Open-weight MoE |\n| Claude 3.5 Sonnet | Anthropic | Balanced | $3/$15 | 8K | Best overall |\n| GPT-4o Mini | OpenAI | Fast | $0.15/$0.60 | 16K | Fallback fast |\n| GPT-4o | OpenAI | Advanced | $5/$15 | 4K | Complex analysis |\n| Kimi K2 Instruct | MoonshotAI | Reasoning | $1/$3 | 131K | Tool use expert |\n| Claude 3 Opus | Anthropic | Reasoning | $15/$75 | 4K | Most capable |\n| Gemini Pro 1.5 | Google | Balanced | $1.25/$5 | 8K | Fallback balanced |\n\n### Debugging Philosophy\n\nThis system is built around the principle of **AI transparency and debuggability**. We use structured data formatting and comprehensive logging to make AI decision-making visible and debuggable.\n\n#### Structured Data with LLML\n\nWe use the `@zenbase/llml` library to convert JavaScript objects into clean, XML-like formatted strings. This approach provides several benefits:\n\n- **Human-readable logs**: Complex webhook payloads become easy to scan and understand\n- **AI-friendly format**: Structured data that AI models can easily parse and reason about\n- **Consistent formatting**: All logged objects follow the same clear structure\n- **Debugging efficiency**: Quickly identify issues in webhook data or AI responses\n\n#### The `stringify-utils.ts` Module\n\nOur formatting utilities provide:\n\n```typescript\n// Convert any object to formatted XML-like string\nconst formatted = stringifyForLogging(webhookPayload);\n\n// Log with clear separators and labels\nlogFormatted(payload, \"Linear Webhook Received\");\n\n// Handle large objects with truncation\nconst truncated = stringifyTruncated(largeObject, 1000);\n```\n\n**Key Functions:**\n- `stringifyForLogging(obj)` - Primary formatter using llml with JSON fallback\n- `logFormatted(obj, label)` - Adds clear separators and labels for log scanning\n- `stringifyTruncated(obj, maxLength)` - Prevents log overflow with large payloads\n\n**Why This Matters:**\n- **AI Debugging**: When AI makes routing decisions, we can see exactly what data it analyzed\n- **Webhook Debugging**: Complex nested webhook payloads become immediately readable\n- **Error Tracking**: Failed AI responses are logged in structured format for analysis\n- **Performance Monitoring**: Easy to spot patterns in notification types and routing decisions\n\nThis structured approach to logging and data formatting makes the entire AI pipeline transparent and debuggable, which is crucial for a system that makes automated decisions about important notifications.\n\n## Project Structure\n\n```\n├── CONFIGURATION.ts # Channel definitions, webhook config, AI prompts, OpenRouter models\n├── ai-service.ts # OpenRouter AI service with multi-model support\n├── ai-model-test.ts # Testing utility for AI models and performance\n├── citation-context.ts # Citation utilities for AI prompting with links\n├── linear-webhook.ts # HTTP webhook handler for Linear (with GET config page)\n├── notification-triage.ts # Shared AI triage system for all notification sources\n├── sentry-webhook.ts # HTTP webhook handler for Sentry (with GET config page)\n├── stringify-utils.ts # LLML-based object formatting utilities for AI debugging\n├── main.tsx # (Reserved for future integrations)\n└── README.md # This file\n```\n\n## Current Features\n\n### ✅ Shared Triage System\n- **Webhook Handler**: Secure endpoint with HMAC-SHA256 signature verification\n- **Configuration Page**: Visit the webhook URL in browser for setup instructions\n- **Event Filtering**: Skips low-value events (views, reads)\n- **Rich Data Processing**: Handles titles, descriptions, assignees, labels, priorities, teams\n- **Async Processing**: Webhooks respond immediately to prevent timeouts, processing happens asynchronously\n\n### ✅ Linear Integration\n- **Error Monitoring**: Processes error events and issue notifications\n- **Severity Mapping**: Converts Sentry levels (fatal, error, warning, etc.) to standard severity\n- **Stack Trace Processing**: Extracts meaningful error context and location information\n- **Environment Awareness**: Routes based on production vs staging environments\n- **Timeout Prevention**: Fast response with background processing to avoid Linear's 4-retry timeout behavior\n\n### ✅ OpenRouter AI Integration\n- **Multi-Model Support**: Access to Claude, GPT-4, Gemini, and other leading AI models\n- **Intelligent Model Selection**: Automatically chooses optimal model based on notification complexity\n- **Cost Optimization**: Uses faster, cheaper models for simple notifications\n- **Fallback Support**: Automatic failover to alternative models if primary fails\n- **Usage Tracking**: Detailed logging of token usage and estimated costs\n- **Configurable Models**: Easy to add new models or adjust selection criteria\n- **Timeout Protection**: 25-second timeout on AI requests to prevent webhook timeouts\n- **Unified Interface**: All webhook sources convert to standardized `NotificationData` format\n- **Citation Context**: Advanced AI prompting with proper link generation and ID replacement\n- **Fallback Logic**: Intelligent routing even when AI fails, based on source and priority\n- **Enhanced Metadata**: Tracks urgency, suggested actions, and formatted summaries\n\n### ✅ Sentry Integration\n- **Link Generation**: Automatic creation of proper URLs and citations in AI responses\n- **ID Replacement**: Converts UUIDs and IDs to readable citation keys for better AI understanding\n- **Structured References**: Maintains mapping between citation keys and actual resources\n- **Markdown Output**: Generates properly formatted links in final notifications\n\n### ✅ Email Delivery System\n- **Multi-Channel Support**: Can notify multiple channels for critical issues\n- **Rich Formatting**: HTML emails with structured information\n- **Branded Messages**: Consistent formatting with clear subject prefixes\n- **Error Handling**: Graceful failure handling per channel\n\n### ✅ Configurable Channel System\nPre-configured channels include:\n- **Engineering Critical**: P0/P1 issues, outages, security incidents\n- **Engineering General**: Feature development, code reviews, technical debt\n- **Product Team**: UX issues, feature requests, business logic\n- **DevOps & Infrastructure**: Deployment, monitoring, performance\n- **Quality Assurance**: Testing, automation, bug reports\n- **General Notifications**: Low-priority and uncategorized items\n\n## Setup Instructions\n\n### 1. Environment Variables\nSet these in your Val Town environment:\n- `LINEAR_WEBHOOK_SECRET` - Your Linear webhook signing secret\n- `SENTRY_WEBHOOK_SECRET` - Your Sentry webhook client secret\n- `OPENROUTER_API_KEY` - Your OpenRouter API key (get from https://openrouter.ai/)\n\n### 2. Configure Channels\nUpdate email addresses in `CONFIGURATION.ts` to match your organization:\n```typescript\n{\n id: 'engineering-critical',\n name: 'Engineering Critical',\n email: 'your-critical-team@company.com', // Update this\n description: '...',\n // ...\n}\n```\n\n### 3. Linear Webhook Setup\n1. Visit your Linear webhook URL in a browser to see configuration instructions\n2. Copy the webhook URL from the configuration page\n3. In Linear: Settings → API → Webhooks → Create webhook\n4. Paste the URL and copy the signing secret\n5. Add the signing secret to Val Town as `LINEAR_WEBHOOK_SECRET`\n6. Select events to monitor (Issues, Comments, Projects recommended)\n\n### 4. Sentry Webhook Setup\n1. Visit your Sentry webhook URL in a browser to see configuration instructions\n2. Copy the webhook URL from the configuration page\n3. In Sentry: Settings → Developer Settings → Internal Integrations\n4. Create new integration or edit existing one\n5. Add the webhook URL and copy the client secret\n6. Add the client secret to Val Town as `SENTRY_WEBHOOK_SECRET`\n7. Enable permissions and subscribe to Error/Issue events\n\n### 5. Testing AI Models\nUse the built-in testing utility to verify your OpenRouter setup:\n\n```typescript\nimport runTests from './ai-model-test.ts';\nawait runTests(); // Test all models and use cases\n```\n\nOr test individual components:\n```typescript\nimport { aiService } from './ai-service.ts';\nimport { testModel, showModelInfo } from './ai-model-test.ts';\n\n// Show available models and pricing\nawait showModelInfo();\n\n// Test a specific model\nawait testModel('anthropic/claude-3.5-sonnet', 'Hello, test message');\n\n// Test notification triage scenario\nawait testTriageScenario();\n```\n\n### 6. Model Configuration\nUpdate model selection in `CONFIGURATION.ts` if needed:\n```typescript\nexport const AI_CONFIG = {\n modelSelection: {\n default: 'anthropic/claude-3.5-sonnet', // Change default model\n fast: 'openai/gpt-oss-20b', // Ultra-cheap for simple notifications \n advanced: 'openai/gpt-4o', // For critical analysis\n reasoning: 'moonshotai/kimi-k2' // For complex scenarios with tool use\n }\n};\n```\n\n**New Models Added:**\n- **GPT-OSS 20B**: Open-weight model with 131K context, extremely cost-effective at $0.10/$0.50 per million tokens\n- **Kimi K2 Instruct**: 1T parameter MoE model optimized for coding, reasoning, and tool use with 131K context\n### 7. Testing\n- **Linear**: Create or update an issue to test the integration\n- **Sentry**: Trigger an error in your application to test the integration\n- **AI Models**: Use `ai-model-test.ts` to verify OpenRouter connectivity\n- Check Val Town logs for debugging information\n\n## How It Works\n\n### Webhook Processing Flow\n1. **Receive**: Linear sends webhook with HMAC signature\n2. **Verify**: Validate signature using signing secret\n3. **Parse**: Extract relevant data (title, description, priority, labels, etc.)\n4. **Respond**: Return 200 OK immediately to prevent timeout (Linear retries 4 times on timeout)\n5. **Analyze**: Send to AI via OpenRouter with intelligent model selection (async)\n6. **Route**: AI selects appropriate channel(s) with reasoning (async)\n7. **Format**: Create tailored message for target audience (async)\n8. **Deliver**: Send email notification(s) (async)\n9. **Log**: Record success/failure, token usage, and costs for debugging (async)\n\n### Timeout Prevention Strategy\nThe system prevents webhook timeouts through several mechanisms:\n- **Immediate Response**: Webhook returns 200 OK within milliseconds\n- **Async Processing**: AI analysis happens in background after response\n- **Fast Default Model**: Uses GPT-4o Mini instead of Claude 3.5 Sonnet for speed\n- **Reduced Token Limits**: 400 tokens max instead of 800 for faster generation\n- **AI Request Timeout**: 25-second timeout on AI calls to prevent hanging\n- **Optimized Model Selection**: Uses fast models for critical issues to ensure quick processing\n\n### AI Model Selection Logic\nThe system automatically chooses the best model for each notification, optimized for speed to prevent timeouts:\n\n```typescript\n// Simple notification → Ultra-fast model (GPT-OSS 20B)\nif (severity === 'low' && !description) useCase = 'fast';\n\n// Critical issue → Fast model (GPT-OSS 20B) - changed from advanced for speed\nelse if (severity === 'critical' || priority === 1) useCase = 'fast';\n\n// Complex scenario → Balanced model (GPT-4o Mini) - changed from reasoning for speed\nelse if (labels.length > 3 || description.length > 500) useCase = 'balanced';\n\n// Default → Balanced model (GPT-4o Mini) - changed from Claude for speed\nelse useCase = 'balanced';\n```\n\n### OpenRouter Provider Configuration\n\nAll requests are configured to use the **Groq** provider through OpenRouter for optimal pricing and performance:\n\n```typescript\nprovider: {\n order: [\"groq\"] // Forces routing through Groq for better prices\n}\n```\n\n### AI Triage Logic\nThe AI considers:\n- **Keywords**: Matches against channel-specific keywords\n- **Priority**: Routes high-priority items to critical channels\n- **Team Context**: Considers team assignments and technical domains\n- **Content Analysis**: Analyzes descriptions for technical vs. product issues\n- **Multi-Channel Logic**: Can notify multiple channels for critical issues\n\n### Channel Selection Criteria\nEach channel has:\n- **Description**: Detailed explanation of what belongs there\n- **Keywords**: Specific terms that indicate relevance\n- **Priority Level**: High/medium/low for urgency-based routing\n- **Email Address**: Delivery target\n\n## Future Roadmap\n\n### 🚧 Planned Integrations\n- **GitHub**: PR reviews, security alerts, deployment status\n- **Slack**: Direct Slack channel delivery (alternative to email)\n- **PagerDuty**: Incident management integration\n- **Custom Webhooks**: Generic webhook handler for other services\n\n### 🚧 Enhanced Features\n- **Smart Scheduling**: Respect time zones and on-call schedules\n- **Escalation Logic**: Auto-escalate if no response within timeframe\n- **Analytics Dashboard**: Track notification patterns and effectiveness\n- **Custom Rules**: User-defined routing rules beyond AI\n- **Digest Mode**: Batch low-priority notifications\n\n## Contributing\n\nWhen adding new integrations:\n1. Create a new webhook handler file (e.g., `sentry-webhook.ts`)\n2. Convert the source payload to `NotificationData` format in your handler\n3. Call `processNotification(notificationData)` from the shared triage system\n4. Include GET endpoint for configuration instructions\n5. Add any source-specific configuration to `CONFIGURATION.ts`\n6. Update this README with new features\n\n### Adding New Webhook Sources\n\nThe shared triage system expects all notifications in this standardized format:\n\n```typescript\ninterface NotificationData {\n source: string; // e.g., 'linear', 'sentry', 'billing'\n id: string; // unique identifier\n type: string; // e.g., 'issue', 'error', 'payment_failed'\n action: string; // e.g., 'created', 'updated', 'resolved'\n title: string;\n description?: string;\n priority?: number | string;\n severity?: 'low' | 'medium' | 'high' | 'critical';\n labels?: string[];\n team?: string;\n assignee?: { name: string; email?: string };\n url?: string;\n // ... additional fields\n}\n```\n\nThe system will automatically:\n- Generate citation contexts for better AI prompting\n- Create proper links in notifications\n- Apply fallback routing logic\n- Format emails consistently across all sources\n\n## Debugging\n\n- **Webhook Logs**: Check Val Town logs for processing details\n- **Structured Logging**: All webhook payloads and AI responses are logged using LLML formatting for easy reading\n- **AI Transparency**: Triage reasoning is logged with clear separators (e.g., \"=== AI Triage Result ===\")\n- **Email Delivery**: Individual channel delivery status is logged\n- **Signature Verification**: Failed authentications are logged with details\n- **LLML Formatting**: Complex objects are automatically formatted as readable XML-like structures\n\n### Example Log Output\n```\n=== Linear Webhook Payload ===\n\n create\n Issue\n \n Critical payment bug\n 1\n \n critical\n \n \n\n=== End Linear Webhook Payload ===\n\n=== AI Triage Result ===\n\n \n engineering-critical\n \n High priority payment issue requires immediate engineering attention\n\n=== End AI Triage Result ===\n```",[],[268],"f27edea0-7c46-11f0-8d37-0224a6c84d84","LINEAR_ISSUE_CREATED_EXAMPLE.txt","d15eb286-7c46-11f0-8785-0224a6c84d84",62,["D",1755530460628],5,"create\n\n bf32e89b-aaf8-4431-9790-8034b94c0988\n Charlie\n 072c476d-9b26-4fba-bcb3-692e798f6485@oauthapp.linear.app\n https://public.linear.app/9f748af2-7258-4540-a068-817e1b0062f8/eae8dc19-fa4e-4c28-9327-ac7e7fb98409/256x256/6ab0f222-0e95-4042-a349-e9fc80b79d47\n https://linear.app/phosphor/profiles/charlie\n user\n\n2025-08-18T15:10:59.609Z\n\n 05f91f23-bce8-4a14-9abe-d965a6e47ff1\n 2025-08-18T15:10:59.609Z\n 2025-08-18T15:10:59.609Z\n 105\n Val Town service: issue notification triage (Receive -> Convert -> Delegate -> Deliver)\n 0\n -11691\n -50714\n 2025-08-18T15:10:59.678Z\n all\n 2025-08-18T15:10:59.683Z\n \n ac9636f1-4ac0-481f-94f7-7efa039defa9\n \n 50636689-fc9d-495b-953b-9b17cf74b90b\n\n bf32e89b-aaf8-4431-9790-8034b94c0988\n a2f2b1db-992c-4325-9994-ce97c6a224d2\n 8a87416e-b6a6-4e73-b925-8e9844e9a421\n\n No priority\n null\n ENG-105\n https://linear.app/phosphor/issue/ENG-105/val-town-service-issue-notification-triage-receive-convert-delegate\n \n bf32e89b-aaf8-4431-9790-8034b94c0988\n a2f2b1db-992c-4325-9994-ce97c6a224d2\n \n \n a2f2b1db-992c-4325-9994-ce97c6a224d2\n Cole Lawrence\n cole@phosphor.co\n https://public.linear.app/773e1976-1cf3-4f20-a94a-8838b9604f8c/a84b0191-0e19-4374-8174-6d15314bd7a4/db0d7b19-b0ee-4c0a-ba31-2fdf62a50cf3\n https://linear.app/phosphor/profiles/cole\n \n \n 8a87416e-b6a6-4e73-b925-8e9844e9a421\n #FC7840\n Triage\n triage\n \n \n 50636689-fc9d-495b-953b-9b17cf74b90b\n ENG\n Engineering\n \n \n \n ac9636f1-4ac0-481f-94f7-7efa039defa9\n #26b5ce\n Exploration\n 2ca86760-7be8-4c9e-8c19-0662a0c209a8\n \n \n \n Build a Val Town–hosted webhook service that normalizes issue notifications, delegates routing to our shared AI triage system, and delivers concise, actionable messages to the right audiences. This should reduce notification fatigue and improve signal/noise.\n \n *Source:* Slack thread [https://phosphor-group.slack.com/archives/C09AZBC6MNY/p1755529682771019](https://phosphor-group.slack.com/archives/C09AZBC6MNY/p1755529682771019) (Aug 18, 2025)\n \n ## Goals\n \n * Cut noisy, broadcast-style alerts; send targeted, actionable notifications\n * Centralize triage logic so sources can be added without rework\n \n ## Scope (pipeline)\n \n 1. **Receive & Validate** – HTTP endpoint(s) on Val Town for source webhooks; verify signatures/headers, basic schema validation\n 2. **Convert** – Map each payload to a standard `NotificationData` structure\n 3. **Delegate** – Call the shared AI triage service for routing decisions (audience, urgency, suppress/merge)\n 4. **Deliver** – Format and send notifications (Slack channels/DMs; room to add email later)\n \n ### Initial sources (MVP)\n \n * GitHub: issues, PRs, comments, and label/state changes\n * Linear: issue created/updated (subset to start)\n \n ### Standard shape (draft)\n \n ```ts\n export type NotificationData = {\n source: 'github' | 'linear';\n kind: 'issue' | 'pr' | 'comment';\n id: string; // source id (node id or number)\n title?: string;\n body?: string;\n url?: string;\n project?: string; // repo or team/project\n labels?: string[];\n assignees?: string[];\n createdAt: string; // ISO\n updatedAt?: string; // ISO\n author?: { name: string; id?: string };\n };\n ```\n \n ### Delivery (Slack)\n \n * Compact, skim-friendly block with: title, summary, key labels/state, primary action link\n * Optional: thread replies for follow-ups or merged events\n \n ### Security & ops\n \n * Validate upstream signatures (e.g., GitHub HMAC), rate-limit, and dedupe\n * Secrets via Val Town environment; minimal request logging with PII scrubbing; 7‑day retention\n * Basic health check endpoint and error alerting\n \n ## Acceptance criteria\n \n * Val Town endpoint accepts GitHub and Linear webhooks and returns 2xx on valid payloads\n * Payloads convert to `NotificationData` and are persisted/transiently queued (in‑memory ok for MVP)\n * Delegation call to AI triage service returns a routing decision and is honored\n * Slack message is posted to the decided destination with links and minimal boilerplate\n * Signature verification and a simple dedupe mechanism are in place\n * README in the Val repo documents setup (secrets, webhook config, sample payloads)\n \n ## Open questions\n \n * Which Slack destinations for MVP (exact channels/DM groups)?\n * Confirm AI triage endpoint + contract (URL, auth, request/response schema)\n * Any non-issue sources to include next (Sentry alerts, CI failures)?\n \n {\"type\":\"doc\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Build a Val Town–hosted webhook service that normalizes issue notifications, delegates routing to our shared AI triage system, and delivers concise, actionable messages to the right audiences. This should reduce notification fatigue and improve signal/noise.\"}]},{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Source:\",\"marks\":[{\"type\":\"em\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" Slack thread \"},{\"type\":\"text\",\"text\":\"https://phosphor-group.slack.com/archives/C09AZBC6MNY/p1755529682771019\",\"marks\":[{\"type\":\"link\",\"attrs\":{\"href\":\"https://phosphor-group.slack.com/archives/C09AZBC6MNY/p1755529682771019\"}}]},{\"type\":\"text\",\"text\":\" (Aug 18, 2025)\"}]},{\"type\":\"heading\",\"attrs\":{\"level\":2,\"id\":\"1baf5da9-7f5f-4bcc-a50c-82bcefb8493a\"},\"content\":[{\"type\":\"text\",\"text\":\"Goals\"}]},{\"type\":\"bullet_list\",\"content\":[{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Cut noisy, broadcast-style alerts; send targeted, actionable notifications\"}]}]},{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Centralize triage logic so sources can be added without rework\"}]}]}]},{\"type\":\"heading\",\"attrs\":{\"level\":2,\"id\":\"6b985aa5-0f42-44ba-b539-0955337c8688\"},\"content\":[{\"type\":\"text\",\"text\":\"Scope (pipeline)\"}]},{\"type\":\"ordered_list\",\"attrs\":{\"order\":1},\"content\":[{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Receive & Validate\",\"marks\":[{\"type\":\"strong\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" – HTTP endpoint(s) on Val Town for source webhooks; verify signatures/headers, basic schema validation\"}]}]},{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Convert\",\"marks\":[{\"type\":\"strong\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" – Map each payload to a standard \"},{\"type\":\"text\",\"text\":\"NotificationData\",\"marks\":[{\"type\":\"code\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" structure\"}]}]},{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Delegate\",\"marks\":[{\"type\":\"strong\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" – Call the shared AI triage service for routing decisions (audience, urgency, suppress/merge)\"}]}]},{\"type\":\"list_item\",\"content\":[{\"type\":\"paragraph\",\"content\":[{\"type\":\"text\",\"text\":\"Deliver\",\"marks\":[{\"type\":\"strong\",\"attrs\":{}}]},{\"type\":\"text\",\"text\":\" – Format and send notifications (Slack channels/DMs; room to ad…",[],[276],"cecd62b2-7f8c-11f0-b8ec-0224a6c84d84","CONFIGURATION.ts","b7df0a40-7c41-11f0-a032-0224a6c84d84",98,["D",1755890319008],14,"/**\n * Configuration for notification triage system\n *\n * This file contains the channel definitions and their descriptions\n * to help AI determine the most appropriate notification targets.\n */\n\nexport interface NotificationChannel {\n id: string;\n name: string;\n email: string;\n description: string;\n keywords: string[];\n priority: \"high\" | \"medium\" | \"low\";\n}\n\n// Using a test notifications email address until we wire up everything.\nconst TEST_EMAIL =\n \"slack-notifications-t-aaaaqsqswnttmuggcrkwq5ltue@phosphor-group.slack.com\";\n\n/**\n * Channel definitions for notification routing\n * Each channel has a description to help AI understand what content belongs there\n */\nexport const NOTIFICATION_CHANNELS: NotificationChannel[] = [\n {\n id: \"eng-general\",\n name: \"Engineering General\",\n email: TEST_EMAIL,\n description:\n \"General engineering discussions, feature development, code reviews, technical debt, and medium-priority issues\",\n keywords: [\"feature\", \"development\", \"code\", \"technical\", \"p2\", \"p3\"],\n priority: \"medium\",\n },\n {\n id: \"references-product-and-design\",\n name: \"Product Team\",\n email: TEST_EMAIL,\n description:\n \"Product-related issues, user experience problems, feature requests, and business logic concerns\",\n keywords: [\"product\", \"user\", \"feature\", \"business\", \"ux\", \"ui\"],\n priority: \"medium\",\n },\n {\n id: \"devops\",\n name: \"DevOps & Infrastructure\",\n email: TEST_EMAIL,\n description:\n \"Infrastructure issues, deployment problems, monitoring alerts, performance issues, and system maintenance\",\n keywords: [\n \"infrastructure\",\n \"deployment\",\n \"monitoring\",\n \"performance\",\n \"devops\",\n \"ci/cd\",\n ],\n priority: \"high\",\n },\n {\n id: \"qa\",\n name: \"Quality Assurance\",\n email: TEST_EMAIL,\n description:\n \"Testing-related issues, quality concerns, bug reports, and test automation problems\",\n keywords: [\"testing\", \"qa\", \"quality\", \"automation\", \"bug\"],\n priority: \"medium\",\n },\n {\n id: \"general\",\n name: \"General Notifications\",\n email: TEST_EMAIL,\n description:\n \"Low-priority notifications, informational updates, and items that don't fit other categories\",\n keywords: [\"info\", \"update\", \"general\"],\n priority: \"low\",\n },\n];\n\n/**\n * Configuration for different webhook sources\n */\nexport const WEBHOOK_CONFIG = {\n linear: {\n signingSecret: Deno.env.get(\"LINEAR_WEBHOOK_SECRET\") || \"\",\n enabled: true,\n },\n sentry: {\n signingSecret: Deno.env.get(\"SENTRY_WEBHOOK_SECRET\") || \"\",\n enabled: true,\n },\n};\n\n/**\n * AI prompt configuration for notification triage\n */\nexport const AI_PROMPTS = {\n channelSelection: `\nYou are a notification triage system. Your job is to analyze incoming notifications and determine the most appropriate channel(s) to notify.\n\nGiven a notification with the following information:\n- Source (e.g., Linear, Sentry)\n- Title/Summary\n- Description/Content\n- Priority/Severity\n- Labels/Tags\n\nAnd the following available channels:\n{channels}\n\nDetermine which channel(s) should receive this notification. Consider:\n1. The urgency and severity of the issue\n2. The technical domain (engineering, product, devops, etc.)\n3. The keywords and descriptions of each channel\n4. Whether multiple channels should be notified\n\nRespond with a JSON object containing:\n- selectedChannels: array of channel IDs\n- reasoning: brief explanation of why these channels were selected\n- summary: a concise summary tailored for the selected audience\n`,\n\n messageFormatting: `\nFormat this notification for delivery to the selected channel. The message should be:\n- Clear and actionable\n- Include relevant links in proper format\n- Tailored to the audience of the channel\n- Include a brief summary of what happened and what action might be needed\n\nNotification data: {notificationData}\nTarget channel: {channelName} - {channelDescription}\n\nFormat the message appropriately for email delivery.\n`,\n};\n\n/**\n * AI Provider Configuration - OpenRouter\n */\nexport interface AIModel {\n id: string;\n name: string;\n provider: string;\n maxTokens: number;\n costPer1kTokens: {\n input: number;\n output: number;\n };\n description: string;\n useCase: \"fast\" | \"balanced\" | \"advanced\" | \"reasoning\";\n}\n\nexport const AI_MODELS: AIModel[] = [\n {\n id: \"anthropic/claude-3.5-sonnet\",\n name: \"Claude 3.5 Sonnet\",\n provider: \"Anthropic\",\n maxTokens: 8192,\n costPer1kTokens: { input: 0.003, output: 0.015 },\n description: \"Best balance of intelligence, speed, and cost for most tasks\",\n useCase: \"balanced\",\n },\n {\n id: \"openai/gpt-4o-mini\",\n name: \"GPT-4o Mini\",\n provider: \"OpenAI\",\n maxTokens: 16384,\n costPer1kTokens: { input: 0.00015, output: 0.0006 },\n description: \"Fast and cost-effective for simple triage tasks\",\n useCase: \"fast\",\n },\n {\n id: \"openai/gpt-4o\",\n name: \"GPT-4o\",\n provider: \"OpenAI\",\n maxTokens: 4096,\n costPer1kTokens: { input: 0.005, output: 0.015 },\n description: \"Advanced reasoning for complex notification analysis\",\n useCase: \"advanced\",\n },\n {\n id: \"anthropic/claude-3-opus\",\n name: \"Claude 3 Opus\",\n provider: \"Anthropic\",\n maxTokens: 4096,\n costPer1kTokens: { input: 0.015, output: 0.075 },\n description: \"Most capable model for complex reasoning and analysis\",\n useCase: \"reasoning\",\n },\n {\n id: \"google/gemini-pro-1.5\",\n name: \"Gemini Pro 1.5\",\n provider: \"Google\",\n maxTokens: 8192,\n costPer1kTokens: { input: 0.00125, output: 0.005 },\n description: \"Good balance of capability and cost\",\n useCase: \"balanced\",\n },\n {\n id: \"openai/gpt-oss-20b\",\n name: \"GPT-OSS 20B\",\n provider: \"OpenAI\",\n maxTokens: 131072,\n costPer1kTokens: { input: 0.0001, output: 0.0005 },\n description: \"Open-weight 21B parameter MoE model with 3.6B active parameters, optimized for low-latency inference and consumer hardware\",\n useCase: \"fast\",\n },\n {\n id: \"moonshotai/kimi-k2\",\n name: \"Kimi K2 Instruct\",\n provider: \"MoonshotAI\",\n maxTokens: 131072,\n costPer1kTokens: { input: 0.001, output: 0.003 },\n description: \"1T parameter MoE model with 32B active parameters, excels at coding, reasoning, and tool use with long-context support\",\n useCase: \"reasoning\",\n },\n];\n\nexport const AI_CONFIG = {\n // OpenRouter configuration\n openRouter: {\n apiKey: Deno.env.get(\"OPENROUTER_API_KEY\") || \"\",\n baseUrl: \"https://openrouter.ai/api/v1\",\n siteName: \"Notification Triage System\",\n siteUrl: \"https://eng.phosphor.co\", // Update this to your actual domain\n },\n\n // Provider configuration - can be overridden per request\n defaultProvider: {\n order: [\"groq\"] // Default to Groq for speed and cost\n },\n \n // Alternative providers for testing\n providers: {\n groq: { order: [\"groq\"] },\n openai: { order: [\"openai\"] },\n anthropic: { order: [\"anthropic\"] },\n auto: {} // Let OpenRouter choose automatically\n },\n\n // Model selection by use case\n modelSelection: {\n // Default model for most triage operations - using fast model to avoid timeouts\n default: \"openai/gpt-4o-mini\",\n\n // Fast model for simple notifications - using the new ultra-cheap GPT-OSS 20B\n fast: \"openai/gpt-oss-20b\",\n\n // Advanced model for complex analysis\n advanced: \"openai/gpt-4o\",\n\n // Reasoning model for difficult edge cases - using the new Kimi K2 for better tool use\n reasoning: \"moonshotai/kimi-k2\",\n },\n\n // Model parameters\n defaultParams: {\n temperature: 0.1,\n maxTokens: 800,\n topP: 1.0,\n },\n\n // Fallback configuration\n fallback: {\n enabled: true,\n // If primary model fails, try these in order (cheapest first)\n models: [\"openai/gpt-oss-20b\", \"openai/gpt-4o-mini\", \"google/gemini-pro-1.5\"],\n },\n};\n\n/**\n * Get AI model configuration by ID\n */\nexport function getModelConfig(modelId: string): AIModel | undefined {\n return AI_MODELS.find((model) => model.id === modelId);\n}\n\n/**\n * Get model ID by use case\n */\nexport function getModelByUseCase(\n useCase: \"fast\" | \"balanced\" | \"advanced\" | \"reasoning\",\n): string {\n // Use the configured model selection rather than searching by useCase property\n switch (useCase) {\n case \"fast\":\n return AI_CONFIG.modelSelection.fast;\n case \"balanced\":\n return AI_CONFIG.modelSelection.default;\n case \"advanced\":\n return AI_CONFIG.modelSelection.advanced;\n case \"reasoning\":\n return AI_CONFIG.modelSelection.reasoning;\n default:\n return AI_CONFIG.modelSelection.default;\n }\n}\n\n/**\n * Email configuration\n */\nexport const EMAIL_CONFIG = {\n fromAddress: TEST_EMAIL,\n fromName: \"Notification Triage System\",\n subjectPrefix: \"[TRIAGE]\",\n};",[],[285],"54bed964-7f8d-11f0-a645-0224a6c84d84","AGENTS.md","54beda72-7f8d-11f0-a645-0224a6c84d84",100,["D",1755890543729],"# AGENTS.md - Slack Notification Triage Webhook System\n\n## Project Overview\n\nThis is an intelligent notification routing system that receives webhooks from various sources (Linear, Sentry, etc.) and uses AI to determine appropriate channels for delivery. The system analyzes notification content using OpenRouter/AI and routes messages to predefined email channels based on context, priority, and team relevance.\n\n## Technology Stack\n\n- **Runtime**: Deno (hosted on Val.Town)\n- **Language**: TypeScript\n- **AI Provider**: OpenRouter (supports multiple models: Claude, GPT-4, Gemini, etc.)\n- **Debugging**: LLML-based structured logging via `@zenbase/llml`\n- **Environment**: Val.Town cloud platform\n\n## Key Commands\n\nSince this runs on Val.Town, there are no traditional build/test commands. Development workflow:\n\n- **Type Check**: Deno handles TypeScript compilation automatically\n- **Lint**: `deno lint` (configured in deno.json)\n- **Test AI Models**: Run `ai-model-test.ts` to verify OpenRouter integration\n- **Deploy**: Save files to Val.Town (auto-deploys)\n\n## Development Workflow\n\n1. **Local Development**: Edit files locally and sync to Val.Town\n2. **Testing**: Use browser to hit webhook URLs for configuration pages\n3. **AI Testing**: Import and run functions from `ai-model-test.ts`\n4. **Debugging**: Check Val.Town logs for structured LLML output\n\n## File Structure\n\n```\n├── CONFIGURATION.ts # Channel definitions, AI config, model settings\n├── ai-service.ts # OpenRouter integration with multi-model support\n├── ai-model-test.ts # AI testing utilities and model verification\n├── citation-context.ts # Citation utilities for AI prompting\n├── linear-webhook.ts # Linear webhook handler with config UI\n├── sentry-webhook.ts # Sentry webhook handler with config UI\n├── notification-triage.ts # Shared AI triage system\n├── stringify-utils.ts # LLML object formatting for debugging\n├── main.tsx # Future integrations (currently empty)\n```\n\n## Environment Variables Required\n\n```\nLINEAR_WEBHOOK_SECRET # Linear webhook signing secret\nSENTRY_WEBHOOK_SECRET # Sentry webhook client secret\nOPENROUTER_API_KEY # OpenRouter API key from https://openrouter.ai/\n```\n\n## Code Style & Conventions\n\n- **TypeScript**: Uses relaxed typing (`noImplicitAny: false`, `strict: false`)\n- **Imports**: ES modules with full URLs for Val.Town compatibility\n- **Error Handling**: Graceful degradation with fallback logic\n- **Logging**: LLML-formatted structured logging via `stringify-utils.ts`\n- **Async**: Heavy use of async/await for webhook processing\n- **Models**: Standardized `NotificationData` interface for all sources\n\n## AI Integration Patterns\n\n- **Model Selection**: Automatic selection based on notification complexity\n- **Timeout Prevention**: 25-second AI request timeout + immediate webhook responses\n- **Cost Optimization**: Uses cheaper models for simple notifications\n- **Fallback Logic**: Graceful routing when AI fails\n- **Citation Context**: Structured prompting with proper link generation\n\n## Testing Approach\n\n```typescript\n// Test AI models\nimport runTests from './ai-model-test.ts';\nawait runTests();\n\n// Test specific model\nimport { testModel } from './ai-model-test.ts';\nawait testModel('anthropic/claude-3.5-sonnet', 'Test message');\n\n// Manual webhook testing: Visit webhook URLs in browser for config\n```\n\n## Debugging Guidelines\n\n- **Structured Logs**: All objects logged using LLML formatting\n- **Clear Separators**: Log sections marked with `=== Section Name ===`\n- **AI Transparency**: Triage decisions logged with reasoning\n- **Webhook Verification**: Signature validation errors are detailed\n- **Token Usage**: AI costs and usage tracked per request\n\n## Performance Considerations\n\n- **Immediate Response**: Webhooks return 200 OK within milliseconds\n- **Async Processing**: AI analysis happens after webhook response\n- **Model Optimization**: Fast models for critical issues to prevent timeouts\n- **Token Limits**: Reduced to 400 tokens max for speed\n- **Provider Routing**: Uses Groq through OpenRouter for better pricing\n\n## Adding New Features\n\n1. **New Webhook Source**: Create handler file, convert to `NotificationData`, use shared triage\n2. **New Channels**: Add to `CONFIGURATION.ts` with description and keywords\n3. **New AI Models**: Update model config in `CONFIGURATION.ts`\n4. **New Integrations**: Extend `main.tsx` for future features\n\n## Security Notes\n\n- **HMAC Verification**: All webhooks verify signatures\n- **Environment Secrets**: Sensitive data in Val.Town environment variables\n- **No Secret Logging**: Redaction markers like `[REDACTED:amp-token]` indicate removed secrets\n",[],[294],["D",1755890283209],[],[18],["D",1755527134608],{"_11":12,"_117":67},{"_11":12,"_113":-5,"_66":67,"_30":65,"_114":115,"_116":24},{"_38":104,"_13":16,"_32":35,"_105":69,"_106":-5,"_107":-5,"_108":-5,"_109":-5,"_110":-5,"_111":14,"_112":12},"ctx",{"_51":-5,"_136":341,"_342":343,"_344":345,"_346":24,"_347":24,"_348":24,"_349":81,"_350":-5,"_351":-7,"_352":-7,"_353":26,"_354":24,"_355":24,"_356":-5,"_357":24,"_358":24,"_359":24,"_360":-5},"ENV",{"_312":313,"_314":315,"_316":16,"_317":318,"_319":320,"_321":322,"_323":324,"_325":326,"_327":328,"_329":330,"_331":332,"_333":334,"_335":336,"_337":338,"_339":340},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","5015b8df46402d8bba4f485ecc37aec202b0e06b","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"]

History · .vtignore