bluera-knowledge 0.9.42 → 0.10.0

package/dist/workers/background-worker-cli.js

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   IntelligentCrawler
-} from "../chunk-MQE32YY6.js";
+} from "../chunk-ITH6FWQY.js";
 import {
   JobService,
   createDocumentId,
@@ -186,7 +186,8 @@ var BackgroundWorker = class {
       const crawlOptions = {
         maxPages: resolvedMaxPages,
         simple: simple ?? false,
-        useHeadless: useHeadless ?? false
+        useHeadless: useHeadless ?? true
+        // Default to headless for reliability
       };
       if (crawlInstruction !== void 0) {
         crawlOptions.crawlInstruction = crawlInstruction;

package/dist/workers/background-worker-cli.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/workers/background-worker.ts","../../src/workers/pid-file.ts","../../src/workers/background-worker-cli.ts"],"sourcesContent":["import { createHash } from 'node:crypto';\nimport { IntelligentCrawler, type CrawlProgress } from '../crawl/intelligent-crawler.js';\nimport { IndexService } from '../services/index.service.js';\nimport { JobService } from '../services/job.service.js';\nimport { StoreService } from '../services/store.service.js';\nimport { createStoreId, createDocumentId } from '../types/brands.js';\nimport type { EmbeddingEngine } from '../db/embeddings.js';\nimport type { LanceStore } from '../db/lance.js';\nimport type { Document } from '../types/document.js';\nimport type { Job } from '../types/job.js';\n\n/**\n * Calculate index progress as a percentage, handling division by zero.\n * @param current - Current number of items processed\n * @param total - Total number of items (may be 0)\n * @param scale - Scale factor for progress (default 100 for 0-100%)\n * @returns Progress value, or 0 if total is 0\n */\nexport function calculateIndexProgress(\n  current: number,\n  total: number,\n  scale: number = 100\n): number {\n  if (total === 0) return 0;\n  return (current / total) * scale;\n}\n\nexport class BackgroundWorker {\n  constructor(\n    private readonly jobService: JobService,\n    private readonly storeService: StoreService,\n    private readonly indexService: IndexService,\n    private readonly lanceStore: LanceStore,\n    private readonly embeddingEngine: EmbeddingEngine\n  ) {}\n\n  /**\n   * Execute a job based on its type\n   */\n  async executeJob(jobId: string): Promise<void> {\n    const job = this.jobService.getJob(jobId);\n\n    if (!job) {\n      throw new Error(`Job ${jobId} not found`);\n    }\n\n    try {\n      // Update to running status\n      this.jobService.updateJob(jobId, {\n        status: 'running',\n        message: `Starting ${job.type} operation...`,\n        progress: 0,\n        details: { startedAt: new Date().toISOString() },\n      });\n\n      // Execute based on job type\n      switch (job.type) {\n        case 'clone':\n          await this.executeCloneJob(job);\n          break;\n        case 'index':\n          await this.executeIndexJob(job);\n          break;\n        case 'crawl':\n          await this.executeCrawlJob(job);\n          break;\n        default:\n          throw new Error(`Unknown job type: ${String(job.type)}`);\n      }\n\n      // Mark as completed\n      this.jobService.updateJob(jobId, {\n        status: 'completed',\n        progress: 100,\n        message: `${job.type} operation completed successfully`,\n        details: { completedAt: new Date().toISOString() },\n      });\n    } catch (error) {\n      // Mark as failed\n      const errorDetails: Record<string, unknown> = {\n        completedAt: new Date().toISOString(),\n      };\n      if (error instanceof Error && error.stack !== undefined) {\n        errorDetails['error'] = error.stack;\n      } else {\n        errorDetails['error'] = String(error);\n      }\n      this.jobService.updateJob(jobId, {\n        status: 'failed',\n        message: error instanceof Error ? error.message : 'Unknown error',\n        details: errorDetails,\n      });\n      throw error;\n    }\n  }\n\n  /**\n   * Execute a clone job (git clone + initial indexing)\n   */\n  private async executeCloneJob(job: Job): Promise<void> {\n    const { storeId } = job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for clone job');\n    }\n\n    // Get the store\n    const store = await this.storeService.get(createStoreId(storeId));\n    if (!store) {\n      throw new Error(`Store ${storeId} not found`);\n    }\n\n    // Clone is already done by the time the job is created\n    // (happens in StoreService.create), so we just need to index\n\n    // Update progress - cloning considered done (30%)\n    this.jobService.updateJob(job.id, {\n      status: 'running',\n      message: 'Repository cloned, starting indexing...',\n      progress: 30,\n    });\n\n    // Index the repository with progress updates\n    const result = await this.indexService.indexStore(\n      store,\n      (event: { type: string; current: number; total: number; message: string }) => {\n        // Check if job was cancelled\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        // Indexing is 70% of total progress (30-100%)\n        const indexProgress = calculateIndexProgress(event.current, event.total, 70);\n        const totalProgress = 30 + indexProgress;\n\n        this.jobService.updateJob(job.id, {\n          message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n          progress: Math.min(99, totalProgress), // Cap at 99 until fully complete\n          details: {\n            filesProcessed: event.current,\n            totalFiles: event.total,\n          },\n        });\n      }\n    );\n\n    if (!result.success) {\n      throw result.error;\n    }\n  }\n\n  /**\n   * Execute an index job (re-indexing existing store)\n   */\n  private async executeIndexJob(job: Job): Promise<void> {\n    const { storeId } = job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for index job');\n    }\n\n    // Get the store\n    const store = await this.storeService.getByIdOrName(createStoreId(storeId));\n    if (!store) {\n      throw new Error(`Store ${storeId} not found`);\n    }\n\n    // Index with progress updates\n    const result = await this.indexService.indexStore(\n      store,\n      (event: { type: string; current: number; total: number; message: string }) => {\n        // Check if job was cancelled\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        const progress = calculateIndexProgress(event.current, event.total);\n\n        this.jobService.updateJob(job.id, {\n          message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n          progress: Math.min(99, progress), // Cap at 99 until fully complete\n          details: {\n            filesProcessed: event.current,\n            totalFiles: event.total,\n          },\n        });\n      }\n    );\n\n    if (!result.success) {\n      throw result.error;\n    }\n  }\n\n  /**\n   * Execute a crawl job (web crawling + indexing)\n   */\n  private async executeCrawlJob(job: Job): Promise<void> {\n    const { storeId, url, crawlInstruction, extractInstruction, maxPages, simple, useHeadless } =\n      job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for crawl job');\n    }\n    if (url === undefined || typeof url !== 'string') {\n      throw new Error('URL required for crawl job');\n    }\n\n    // Get the store\n    const store = await this.storeService.get(createStoreId(storeId));\n    if (store?.type !== 'web') {\n      throw new Error(`Web store ${storeId} not found`);\n    }\n\n    const resolvedMaxPages = typeof maxPages === 'number' ? maxPages : 50;\n    const crawler = new IntelligentCrawler();\n\n    // Listen for progress events\n    crawler.on('progress', (progress: CrawlProgress) => {\n      // Check if job was cancelled - just return early, for-await loop will throw and finally will cleanup\n      const currentJob = this.jobService.getJob(job.id);\n      if (currentJob?.status === 'cancelled') {\n        return;\n      }\n\n      // Crawling is 80% of total progress (0-80%)\n      const crawlProgress = (progress.pagesVisited / resolvedMaxPages) * 80;\n\n      this.jobService.updateJob(job.id, {\n        message:\n          progress.message ??\n          `Crawling page ${String(progress.pagesVisited)}/${String(resolvedMaxPages)}`,\n        progress: Math.min(80, crawlProgress),\n        details: { pagesCrawled: progress.pagesVisited },\n      });\n    });\n\n    try {\n      await this.lanceStore.initialize(store.id);\n      const docs: Document[] = [];\n\n      // Build crawl options, only including defined values\n      const crawlOptions: {\n        maxPages: number;\n        simple: boolean;\n        useHeadless: boolean;\n        crawlInstruction?: string;\n        extractInstruction?: string;\n      } = {\n        maxPages: resolvedMaxPages,\n        simple: simple ?? false,\n        useHeadless: useHeadless ?? false,\n      };\n      if (crawlInstruction !== undefined) {\n        crawlOptions.crawlInstruction = crawlInstruction;\n      }\n      if (extractInstruction !== undefined) {\n        crawlOptions.extractInstruction = extractInstruction;\n      }\n\n      // Crawl pages using IntelligentCrawler\n      for await (const result of crawler.crawl(url, crawlOptions)) {\n        // Check cancellation between pages\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        // Embed and index the content (use extracted if available, otherwise markdown)\n        const contentToEmbed = result.extracted ?? result.markdown;\n        const vector = await this.embeddingEngine.embed(contentToEmbed);\n\n        docs.push({\n          id: createDocumentId(`${store.id}-${createHash('md5').update(result.url).digest('hex')}`),\n          content: contentToEmbed,\n          vector,\n          metadata: {\n            type: 'web',\n            storeId: store.id,\n            url: result.url,\n            title: result.title,\n            extracted: result.extracted !== undefined,\n            depth: result.depth,\n            indexedAt: new Date(),\n          },\n        });\n      }\n\n      // Index all documents (remaining 20%)\n      if (docs.length > 0) {\n        this.jobService.updateJob(job.id, {\n          message: 'Indexing crawled documents...',\n          progress: 85,\n        });\n\n        await this.lanceStore.addDocuments(store.id, docs);\n        // Create FTS index for full-text search\n        await this.lanceStore.createFtsIndex(store.id);\n      }\n\n      this.jobService.updateJob(job.id, {\n        message: `Crawled and indexed ${String(docs.length)} pages`,\n        progress: 100,\n        details: { pagesCrawled: docs.length },\n      });\n    } finally {\n      await crawler.stop();\n    }\n  }\n}\n","import fs from 'fs';\nimport path from 'path';\n\n/**\n * Result of a PID file delete operation.\n * Delete operations are best-effort and should not throw.\n */\nexport interface PidFileResult {\n  success: boolean;\n  error?: Error;\n}\n\n/**\n * Context for PID file deletion - indicates when the delete is happening.\n * Used for logging/debugging purposes.\n */\nexport type PidFileDeleteContext = 'sigterm' | 'success' | 'failure';\n\n/**\n * Write PID file - CRITICAL operation that must succeed.\n *\n * If the PID file cannot be written, the job cannot be cancelled through\n * the job management system. This is a critical failure and the job\n * should not proceed.\n *\n * @param pidFile - Absolute path to the PID file\n * @param pid - Process ID to write\n * @throws Error if PID file cannot be written\n */\nexport function writePidFile(pidFile: string, pid: number): void {\n  try {\n    fs.writeFileSync(pidFile, pid.toString(), 'utf-8');\n  } catch (error) {\n    const message = error instanceof Error ? error.message : String(error);\n    throw new Error(\n      `CRITICAL: Failed to write PID file ${pidFile}. ` +\n        `Job cannot be cancelled without PID file. ` +\n        `Original error: ${message}`\n    );\n  }\n}\n\n/**\n * Delete PID file - best-effort cleanup during shutdown.\n *\n * This operation should NEVER throw. During process shutdown (SIGTERM,\n * job success, job failure), failing to delete a PID file should not\n * prevent the process from exiting cleanly.\n *\n * Stale PID files are cleaned up by JobService.cleanupOldJobs().\n *\n * @param pidFile - Absolute path to the PID file\n * @param _context - Context indicating when the delete is happening (for future logging)\n * @returns Result indicating success or failure with error details\n */\nexport function deletePidFile(pidFile: string, _context: PidFileDeleteContext): PidFileResult {\n  try {\n    fs.unlinkSync(pidFile);\n    return { success: true };\n  } catch (error) {\n    // ENOENT = file doesn't exist - that's success (nothing to delete)\n    if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {\n      return { success: true };\n    }\n    // Any other error = failure (permission denied, etc.)\n    return {\n      success: false,\n      error: error instanceof Error ? error : new Error(String(error)),\n    };\n  }\n}\n\n/**\n * Build the path to a PID file for a given job.\n *\n * @param jobsDir - Directory where job files are stored\n * @param jobId - Job identifier\n * @returns Absolute path to the PID file\n */\nexport function buildPidFilePath(jobsDir: string, jobId: string): string {\n  return path.join(jobsDir, `${jobId}.pid`);\n}\n","#!/usr/bin/env node\nimport { BackgroundWorker } from './background-worker.js';\nimport { writePidFile, deletePidFile, buildPidFilePath } from './pid-file.js';\nimport { createServices } from '../services/index.js';\nimport { JobService } from '../services/job.service.js';\n\n/**\n * Background worker CLI entry point\n *\n * Usage: background-worker-cli <job-id>\n *\n * This process runs detached from the parent and executes a single job.\n */\n\nasync function main(): Promise<void> {\n  const jobId = process.argv[2];\n  const dataDir = process.env['BLUERA_DATA_DIR'];\n\n  if (jobId === undefined || jobId === '') {\n    console.error('Error: Job ID required');\n    console.error('Usage: background-worker-cli <job-id>');\n    process.exit(1);\n  }\n\n  // Initialize services\n  const jobService = new JobService(dataDir);\n  const services = await createServices(undefined, dataDir);\n\n  // Write PID file for job cancellation - CRITICAL: must succeed or job cannot be cancelled\n  const pidFile = buildPidFilePath(\n    jobService['jobsDir'], // Access private field for PID path\n    jobId\n  );\n\n  try {\n    writePidFile(pidFile, process.pid);\n  } catch (error) {\n    // CRITICAL: Cannot proceed without PID file - job would be uncancellable\n    console.error(error instanceof Error ? error.message : String(error));\n    process.exit(1);\n  }\n\n  // Handle SIGTERM for graceful shutdown\n  process.on('SIGTERM', () => {\n    console.log(`[${jobId}] Received SIGTERM, cancelling job...`);\n    jobService.updateJob(jobId, {\n      status: 'cancelled',\n      message: 'Job cancelled by user',\n    });\n\n    // Clean up PID file (best-effort - don't block shutdown)\n    const deleteResult = deletePidFile(pidFile, 'sigterm');\n    if (!deleteResult.success && deleteResult.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file during SIGTERM: ${deleteResult.error.message}`\n      );\n    }\n\n    process.exit(0);\n  });\n\n  // Create worker and execute job\n  const worker = new BackgroundWorker(\n    jobService,\n    services.store,\n    services.index,\n    services.lance,\n    services.embeddings\n  );\n\n  try {\n    await worker.executeJob(jobId);\n\n    // Clean up PID file on success (best-effort - don't change exit code)\n    const successCleanup = deletePidFile(pidFile, 'success');\n    if (!successCleanup.success && successCleanup.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file after success: ${successCleanup.error.message}`\n      );\n    }\n\n    console.log(`[${jobId}] Job completed successfully`);\n    process.exit(0);\n  } catch (error) {\n    // Job service already updated with failure status in BackgroundWorker\n    console.error(`[${jobId}] Job failed:`, error);\n\n    // Clean up PID file on failure (best-effort - exit code reflects job failure)\n    const failureCleanup = deletePidFile(pidFile, 'failure');\n    if (!failureCleanup.success && failureCleanup.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file after failure: ${failureCleanup.error.message}`\n      );\n    }\n\n    process.exit(1);\n  }\n}\n\nmain().catch((error: unknown) => {\n  console.error('Fatal error in background worker:', error);\n  process.exit(1);\n});\n"],"mappings":";;;;;;;;;;;;;AAAA,SAAS,kBAAkB;AAkBpB,SAAS,uBACd,SACA,OACA,QAAgB,KACR;AACR,MAAI,UAAU,EAAG,QAAO;AACxB,SAAQ,UAAU,QAAS;AAC7B;AAEO,IAAM,mBAAN,MAAuB;AAAA,EAC5B,YACmB,YACA,cACA,cACA,YACA,iBACjB;AALiB;AACA;AACA;AACA;AACA;AAAA,EAChB;AAAA;AAAA;AAAA;AAAA,EAKH,MAAM,WAAW,OAA8B;AAC7C,UAAM,MAAM,KAAK,WAAW,OAAO,KAAK;AAExC,QAAI,CAAC,KAAK;AACR,YAAM,IAAI,MAAM,OAAO,KAAK,YAAY;AAAA,IAC1C;AAEA,QAAI;AAEF,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,YAAY,IAAI,IAAI;AAAA,QAC7B,UAAU;AAAA,QACV,SAAS,EAAE,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACjD,CAAC;AAGD,cAAQ,IAAI,MAAM;AAAA,QAChB,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF;AACE,gBAAM,IAAI,MAAM,qBAAqB,OAAO,IAAI,IAAI,CAAC,EAAE;AAAA,MAC3D;AAGA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,UAAU;AAAA,QACV,SAAS,GAAG,IAAI,IAAI;AAAA,QACpB,SAAS,EAAE,cAAa,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACnD,CAAC;AAAA,IACH,SAAS,OAAO;AAEd,YAAM,eAAwC;AAAA,QAC5C,cAAa,oBAAI,KAAK,GAAE,YAAY;AAAA,MACtC;AACA,UAAI,iBAAiB,SAAS,MAAM,UAAU,QAAW;AACvD,qBAAa,OAAO,IAAI,MAAM;AAAA,MAChC,OAAO;AACL,qBAAa,OAAO,IAAI,OAAO,KAAK;AAAA,MACtC;AACA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,iBAAiB,QAAQ,MAAM,UAAU;AAAA,QAClD,SAAS;AAAA,MACX,CAAC;AACD,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAMA,SAAK,WAAW,UAAU,IAAI,IAAI;AAAA,MAChC,QAAQ;AAAA,MACR,SAAS;AAAA,MACT,UAAU;AAAA,IACZ,CAAC;AAGD,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,gBAAgB,uBAAuB,MAAM,SAAS,MAAM,OAAO,EAAE;AAC3E,cAAM,gBAAgB,KAAK;AAE3B,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA;AAAA,UACpC,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,cAAc,cAAc,OAAO,CAAC;AAC1E,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAGA,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAEA,cAAM,WAAW,uBAAuB,MAAM,SAAS,MAAM,KAAK;AAElE,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,QAAQ;AAAA;AAAA,UAC/B,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,SAAS,KAAK,kBAAkB,oBAAoB,UAAU,QAAQ,YAAY,IACxF,IAAI;AAEN,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AACA,QAAI,QAAQ,UAAa,OAAO,QAAQ,UAAU;AAChD,YAAM,IAAI,MAAM,4BAA4B;AAAA,IAC9C;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,OAAO,SAAS,OAAO;AACzB,YAAM,IAAI,MAAM,aAAa,OAAO,YAAY;AAAA,IAClD;AAEA,UAAM,mBAAmB,OAAO,aAAa,WAAW,WAAW;AACnE,UAAM,UAAU,IAAI,mBAAmB;AAGvC,YAAQ,GAAG,YAAY,CAAC,aAA4B;AAElD,YAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,UAAI,YAAY,WAAW,aAAa;AACtC;AAAA,MACF;AAGA,YAAM,gBAAiB,SAAS,eAAe,mBAAoB;AAEnE,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SACE,SAAS,WACT,iBAAiB,OAAO,SAAS,YAAY,CAAC,IAAI,OAAO,gBAAgB,CAAC;AAAA,QAC5E,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA,QACpC,SAAS,EAAE,cAAc,SAAS,aAAa;AAAA,MACjD,CAAC;AAAA,IACH,CAAC;AAED,QAAI;AACF,YAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AACzC,YAAM,OAAmB,CAAC;AAG1B,YAAM,eAMF;AAAA,QACF,UAAU;AAAA,QACV,QAAQ,UAAU;AAAA,QAClB,aAAa,eAAe;AAAA,MAC9B;AACA,UAAI,qBAAqB,QAAW;AAClC,qBAAa,mBAAmB;AAAA,MAClC;AACA,UAAI,uBAAuB,QAAW;AACpC,qBAAa,qBAAqB;AAAA,MACpC;AAGA,uBAAiB,UAAU,QAAQ,MAAM,KAAK,YAAY,GAAG;AAE3D,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,iBAAiB,OAAO,aAAa,OAAO;AAClD,cAAM,SAAS,MAAM,KAAK,gBAAgB,MAAM,cAAc;AAE9D,aAAK,KAAK;AAAA,UACR,IAAI,iBAAiB,GAAG,MAAM,EAAE,IAAI,WAAW,KAAK,EAAE,OAAO,OAAO,GAAG,EAAE,OAAO,KAAK,CAAC,EAAE;AAAA,UACxF,SAAS;AAAA,UACT;AAAA,UACA,UAAU;AAAA,YACR,MAAM;AAAA,YACN,SAAS,MAAM;AAAA,YACf,KAAK,OAAO;AAAA,YACZ,OAAO,OAAO;AAAA,YACd,WAAW,OAAO,cAAc;AAAA,YAChC,OAAO,OAAO;AAAA,YACd,WAAW,oBAAI,KAAK;AAAA,UACtB;AAAA,QACF,CAAC;AAAA,MACH;AAGA,UAAI,KAAK,SAAS,GAAG;AACnB,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS;AAAA,UACT,UAAU;AAAA,QACZ,CAAC;AAED,cAAM,KAAK,WAAW,aAAa,MAAM,IAAI,IAAI;AAEjD,cAAM,KAAK,WAAW,eAAe,MAAM,EAAE;AAAA,MAC/C;AAEA,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SAAS,uBAAuB,OAAO,KAAK,MAAM,CAAC;AAAA,QACnD,UAAU;AAAA,QACV,SAAS,EAAE,cAAc,KAAK,OAAO;AAAA,MACvC,CAAC;AAAA,IACH,UAAE;AACA,YAAM,QAAQ,KAAK;AAAA,IACrB;AAAA,EACF;AACF;;;ACvTA,OAAO,QAAQ;AACf,OAAO,UAAU;AA4BV,SAAS,aAAa,SAAiB,KAAmB;AAC/D,MAAI;AACF,OAAG,cAAc,SAAS,IAAI,SAAS,GAAG,OAAO;AAAA,EACnD,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,IAAI;AAAA,MACR,sCAAsC,OAAO,+DAExB,OAAO;AAAA,IAC9B;AAAA,EACF;AACF;AAeO,SAAS,cAAc,SAAiB,UAA+C;AAC5F,MAAI;AACF,OAAG,WAAW,OAAO;AACrB,WAAO,EAAE,SAAS,KAAK;AAAA,EACzB,SAAS,OAAO;AAEd,QAAI,iBAAiB,SAAS,UAAU,SAAS,MAAM,SAAS,UAAU;AACxE,aAAO,EAAE,SAAS,KAAK;AAAA,IACzB;AAEA,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;AAAA,IACjE;AAAA,EACF;AACF;AASO,SAAS,iBAAiB,SAAiB,OAAuB;AACvE,SAAO,KAAK,KAAK,SAAS,GAAG,KAAK,MAAM;AAC1C;;;ACnEA,eAAe,OAAsB;AACnC,QAAM,QAAQ,QAAQ,KAAK,CAAC;AAC5B,QAAM,UAAU,QAAQ,IAAI,iBAAiB;AAE7C,MAAI,UAAU,UAAa,UAAU,IAAI;AACvC,YAAQ,MAAM,wBAAwB;AACtC,YAAQ,MAAM,uCAAuC;AACrD,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,QAAM,aAAa,IAAI,WAAW,OAAO;AACzC,QAAM,WAAW,MAAM,eAAe,QAAW,OAAO;AAGxD,QAAM,UAAU;AAAA,IACd,WAAW,SAAS;AAAA;AAAA,IACpB;AAAA,EACF;AAEA,MAAI;AACF,iBAAa,SAAS,QAAQ,GAAG;AAAA,EACnC,SAAS,OAAO;AAEd,YAAQ,MAAM,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AACpE,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,UAAQ,GAAG,WAAW,MAAM;AAC1B,YAAQ,IAAI,IAAI,KAAK,uCAAuC;AAC5D,eAAW,UAAU,OAAO;AAAA,MAC1B,QAAQ;AAAA,MACR,SAAS;AAAA,IACX,CAAC;AAGD,UAAM,eAAe,cAAc,SAAS,SAAS;AACrD,QAAI,CAAC,aAAa,WAAW,aAAa,UAAU,QAAW;AAC7D,cAAQ;AAAA,QACN,sDAAsD,aAAa,MAAM,OAAO;AAAA,MAClF;AAAA,IACF;AAEA,YAAQ,KAAK,CAAC;AAAA,EAChB,CAAC;AAGD,QAAM,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,EACX;AAEA,MAAI;AACF,UAAM,OAAO,WAAW,KAAK;AAG7B,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,cAAQ;AAAA,QACN,qDAAqD,eAAe,MAAM,OAAO;AAAA,MACnF;AAAA,IACF;AAEA,YAAQ,IAAI,IAAI,KAAK,8BAA8B;AACnD,YAAQ,KAAK,CAAC;AAAA,EAChB,SAAS,OAAO;AAEd,YAAQ,MAAM,IAAI,KAAK,iBAAiB,KAAK;AAG7C,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,cAAQ;AAAA,QACN,qDAAqD,eAAe,MAAM,OAAO;AAAA,MACnF;AAAA,IACF;AAEA,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF;AAEA,KAAK,EAAE,MAAM,CAAC,UAAmB;AAC/B,UAAQ,MAAM,qCAAqC,KAAK;AACxD,UAAQ,KAAK,CAAC;AAChB,CAAC;","names":[]}
+{"version":3,"sources":["../../src/workers/background-worker.ts","../../src/workers/pid-file.ts","../../src/workers/background-worker-cli.ts"],"sourcesContent":["import { createHash } from 'node:crypto';\nimport { IntelligentCrawler, type CrawlProgress } from '../crawl/intelligent-crawler.js';\nimport { IndexService } from '../services/index.service.js';\nimport { JobService } from '../services/job.service.js';\nimport { StoreService } from '../services/store.service.js';\nimport { createStoreId, createDocumentId } from '../types/brands.js';\nimport type { EmbeddingEngine } from '../db/embeddings.js';\nimport type { LanceStore } from '../db/lance.js';\nimport type { Document } from '../types/document.js';\nimport type { Job } from '../types/job.js';\n\n/**\n * Calculate index progress as a percentage, handling division by zero.\n * @param current - Current number of items processed\n * @param total - Total number of items (may be 0)\n * @param scale - Scale factor for progress (default 100 for 0-100%)\n * @returns Progress value, or 0 if total is 0\n */\nexport function calculateIndexProgress(\n  current: number,\n  total: number,\n  scale: number = 100\n): number {\n  if (total === 0) return 0;\n  return (current / total) * scale;\n}\n\nexport class BackgroundWorker {\n  constructor(\n    private readonly jobService: JobService,\n    private readonly storeService: StoreService,\n    private readonly indexService: IndexService,\n    private readonly lanceStore: LanceStore,\n    private readonly embeddingEngine: EmbeddingEngine\n  ) {}\n\n  /**\n   * Execute a job based on its type\n   */\n  async executeJob(jobId: string): Promise<void> {\n    const job = this.jobService.getJob(jobId);\n\n    if (!job) {\n      throw new Error(`Job ${jobId} not found`);\n    }\n\n    try {\n      // Update to running status\n      this.jobService.updateJob(jobId, {\n        status: 'running',\n        message: `Starting ${job.type} operation...`,\n        progress: 0,\n        details: { startedAt: new Date().toISOString() },\n      });\n\n      // Execute based on job type\n      switch (job.type) {\n        case 'clone':\n          await this.executeCloneJob(job);\n          break;\n        case 'index':\n          await this.executeIndexJob(job);\n          break;\n        case 'crawl':\n          await this.executeCrawlJob(job);\n          break;\n        default:\n          throw new Error(`Unknown job type: ${String(job.type)}`);\n      }\n\n      // Mark as completed\n      this.jobService.updateJob(jobId, {\n        status: 'completed',\n        progress: 100,\n        message: `${job.type} operation completed successfully`,\n        details: { completedAt: new Date().toISOString() },\n      });\n    } catch (error) {\n      // Mark as failed\n      const errorDetails: Record<string, unknown> = {\n        completedAt: new Date().toISOString(),\n      };\n      if (error instanceof Error && error.stack !== undefined) {\n        errorDetails['error'] = error.stack;\n      } else {\n        errorDetails['error'] = String(error);\n      }\n      this.jobService.updateJob(jobId, {\n        status: 'failed',\n        message: error instanceof Error ? error.message : 'Unknown error',\n        details: errorDetails,\n      });\n      throw error;\n    }\n  }\n\n  /**\n   * Execute a clone job (git clone + initial indexing)\n   */\n  private async executeCloneJob(job: Job): Promise<void> {\n    const { storeId } = job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for clone job');\n    }\n\n    // Get the store\n    const store = await this.storeService.get(createStoreId(storeId));\n    if (!store) {\n      throw new Error(`Store ${storeId} not found`);\n    }\n\n    // Clone is already done by the time the job is created\n    // (happens in StoreService.create), so we just need to index\n\n    // Update progress - cloning considered done (30%)\n    this.jobService.updateJob(job.id, {\n      status: 'running',\n      message: 'Repository cloned, starting indexing...',\n      progress: 30,\n    });\n\n    // Index the repository with progress updates\n    const result = await this.indexService.indexStore(\n      store,\n      (event: { type: string; current: number; total: number; message: string }) => {\n        // Check if job was cancelled\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        // Indexing is 70% of total progress (30-100%)\n        const indexProgress = calculateIndexProgress(event.current, event.total, 70);\n        const totalProgress = 30 + indexProgress;\n\n        this.jobService.updateJob(job.id, {\n          message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n          progress: Math.min(99, totalProgress), // Cap at 99 until fully complete\n          details: {\n            filesProcessed: event.current,\n            totalFiles: event.total,\n          },\n        });\n      }\n    );\n\n    if (!result.success) {\n      throw result.error;\n    }\n  }\n\n  /**\n   * Execute an index job (re-indexing existing store)\n   */\n  private async executeIndexJob(job: Job): Promise<void> {\n    const { storeId } = job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for index job');\n    }\n\n    // Get the store\n    const store = await this.storeService.getByIdOrName(createStoreId(storeId));\n    if (!store) {\n      throw new Error(`Store ${storeId} not found`);\n    }\n\n    // Index with progress updates\n    const result = await this.indexService.indexStore(\n      store,\n      (event: { type: string; current: number; total: number; message: string }) => {\n        // Check if job was cancelled\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        const progress = calculateIndexProgress(event.current, event.total);\n\n        this.jobService.updateJob(job.id, {\n          message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n          progress: Math.min(99, progress), // Cap at 99 until fully complete\n          details: {\n            filesProcessed: event.current,\n            totalFiles: event.total,\n          },\n        });\n      }\n    );\n\n    if (!result.success) {\n      throw result.error;\n    }\n  }\n\n  /**\n   * Execute a crawl job (web crawling + indexing)\n   */\n  private async executeCrawlJob(job: Job): Promise<void> {\n    const { storeId, url, crawlInstruction, extractInstruction, maxPages, simple, useHeadless } =\n      job.details;\n\n    if (storeId === undefined || typeof storeId !== 'string') {\n      throw new Error('Store ID required for crawl job');\n    }\n    if (url === undefined || typeof url !== 'string') {\n      throw new Error('URL required for crawl job');\n    }\n\n    // Get the store\n    const store = await this.storeService.get(createStoreId(storeId));\n    if (store?.type !== 'web') {\n      throw new Error(`Web store ${storeId} not found`);\n    }\n\n    const resolvedMaxPages = typeof maxPages === 'number' ? maxPages : 50;\n    const crawler = new IntelligentCrawler();\n\n    // Listen for progress events\n    crawler.on('progress', (progress: CrawlProgress) => {\n      // Check if job was cancelled - just return early, for-await loop will throw and finally will cleanup\n      const currentJob = this.jobService.getJob(job.id);\n      if (currentJob?.status === 'cancelled') {\n        return;\n      }\n\n      // Crawling is 80% of total progress (0-80%)\n      const crawlProgress = (progress.pagesVisited / resolvedMaxPages) * 80;\n\n      this.jobService.updateJob(job.id, {\n        message:\n          progress.message ??\n          `Crawling page ${String(progress.pagesVisited)}/${String(resolvedMaxPages)}`,\n        progress: Math.min(80, crawlProgress),\n        details: { pagesCrawled: progress.pagesVisited },\n      });\n    });\n\n    try {\n      await this.lanceStore.initialize(store.id);\n      const docs: Document[] = [];\n\n      // Build crawl options, only including defined values\n      const crawlOptions: {\n        maxPages: number;\n        simple: boolean;\n        useHeadless: boolean;\n        crawlInstruction?: string;\n        extractInstruction?: string;\n      } = {\n        maxPages: resolvedMaxPages,\n        simple: simple ?? false,\n        useHeadless: useHeadless ?? true, // Default to headless for reliability\n      };\n      if (crawlInstruction !== undefined) {\n        crawlOptions.crawlInstruction = crawlInstruction;\n      }\n      if (extractInstruction !== undefined) {\n        crawlOptions.extractInstruction = extractInstruction;\n      }\n\n      // Crawl pages using IntelligentCrawler\n      for await (const result of crawler.crawl(url, crawlOptions)) {\n        // Check cancellation between pages\n        const currentJob = this.jobService.getJob(job.id);\n        if (currentJob?.status === 'cancelled') {\n          throw new Error('Job cancelled by user');\n        }\n\n        // Embed and index the content (use extracted if available, otherwise markdown)\n        const contentToEmbed = result.extracted ?? result.markdown;\n        const vector = await this.embeddingEngine.embed(contentToEmbed);\n\n        docs.push({\n          id: createDocumentId(`${store.id}-${createHash('md5').update(result.url).digest('hex')}`),\n          content: contentToEmbed,\n          vector,\n          metadata: {\n            type: 'web',\n            storeId: store.id,\n            url: result.url,\n            title: result.title,\n            extracted: result.extracted !== undefined,\n            depth: result.depth,\n            indexedAt: new Date(),\n          },\n        });\n      }\n\n      // Index all documents (remaining 20%)\n      if (docs.length > 0) {\n        this.jobService.updateJob(job.id, {\n          message: 'Indexing crawled documents...',\n          progress: 85,\n        });\n\n        await this.lanceStore.addDocuments(store.id, docs);\n        // Create FTS index for full-text search\n        await this.lanceStore.createFtsIndex(store.id);\n      }\n\n      this.jobService.updateJob(job.id, {\n        message: `Crawled and indexed ${String(docs.length)} pages`,\n        progress: 100,\n        details: { pagesCrawled: docs.length },\n      });\n    } finally {\n      await crawler.stop();\n    }\n  }\n}\n","import fs from 'fs';\nimport path from 'path';\n\n/**\n * Result of a PID file delete operation.\n * Delete operations are best-effort and should not throw.\n */\nexport interface PidFileResult {\n  success: boolean;\n  error?: Error;\n}\n\n/**\n * Context for PID file deletion - indicates when the delete is happening.\n * Used for logging/debugging purposes.\n */\nexport type PidFileDeleteContext = 'sigterm' | 'success' | 'failure';\n\n/**\n * Write PID file - CRITICAL operation that must succeed.\n *\n * If the PID file cannot be written, the job cannot be cancelled through\n * the job management system. This is a critical failure and the job\n * should not proceed.\n *\n * @param pidFile - Absolute path to the PID file\n * @param pid - Process ID to write\n * @throws Error if PID file cannot be written\n */\nexport function writePidFile(pidFile: string, pid: number): void {\n  try {\n    fs.writeFileSync(pidFile, pid.toString(), 'utf-8');\n  } catch (error) {\n    const message = error instanceof Error ? error.message : String(error);\n    throw new Error(\n      `CRITICAL: Failed to write PID file ${pidFile}. ` +\n        `Job cannot be cancelled without PID file. ` +\n        `Original error: ${message}`\n    );\n  }\n}\n\n/**\n * Delete PID file - best-effort cleanup during shutdown.\n *\n * This operation should NEVER throw. During process shutdown (SIGTERM,\n * job success, job failure), failing to delete a PID file should not\n * prevent the process from exiting cleanly.\n *\n * Stale PID files are cleaned up by JobService.cleanupOldJobs().\n *\n * @param pidFile - Absolute path to the PID file\n * @param _context - Context indicating when the delete is happening (for future logging)\n * @returns Result indicating success or failure with error details\n */\nexport function deletePidFile(pidFile: string, _context: PidFileDeleteContext): PidFileResult {\n  try {\n    fs.unlinkSync(pidFile);\n    return { success: true };\n  } catch (error) {\n    // ENOENT = file doesn't exist - that's success (nothing to delete)\n    if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {\n      return { success: true };\n    }\n    // Any other error = failure (permission denied, etc.)\n    return {\n      success: false,\n      error: error instanceof Error ? error : new Error(String(error)),\n    };\n  }\n}\n\n/**\n * Build the path to a PID file for a given job.\n *\n * @param jobsDir - Directory where job files are stored\n * @param jobId - Job identifier\n * @returns Absolute path to the PID file\n */\nexport function buildPidFilePath(jobsDir: string, jobId: string): string {\n  return path.join(jobsDir, `${jobId}.pid`);\n}\n","#!/usr/bin/env node\nimport { BackgroundWorker } from './background-worker.js';\nimport { writePidFile, deletePidFile, buildPidFilePath } from './pid-file.js';\nimport { createServices } from '../services/index.js';\nimport { JobService } from '../services/job.service.js';\n\n/**\n * Background worker CLI entry point\n *\n * Usage: background-worker-cli <job-id>\n *\n * This process runs detached from the parent and executes a single job.\n */\n\nasync function main(): Promise<void> {\n  const jobId = process.argv[2];\n  const dataDir = process.env['BLUERA_DATA_DIR'];\n\n  if (jobId === undefined || jobId === '') {\n    console.error('Error: Job ID required');\n    console.error('Usage: background-worker-cli <job-id>');\n    process.exit(1);\n  }\n\n  // Initialize services\n  const jobService = new JobService(dataDir);\n  const services = await createServices(undefined, dataDir);\n\n  // Write PID file for job cancellation - CRITICAL: must succeed or job cannot be cancelled\n  const pidFile = buildPidFilePath(\n    jobService['jobsDir'], // Access private field for PID path\n    jobId\n  );\n\n  try {\n    writePidFile(pidFile, process.pid);\n  } catch (error) {\n    // CRITICAL: Cannot proceed without PID file - job would be uncancellable\n    console.error(error instanceof Error ? error.message : String(error));\n    process.exit(1);\n  }\n\n  // Handle SIGTERM for graceful shutdown\n  process.on('SIGTERM', () => {\n    console.log(`[${jobId}] Received SIGTERM, cancelling job...`);\n    jobService.updateJob(jobId, {\n      status: 'cancelled',\n      message: 'Job cancelled by user',\n    });\n\n    // Clean up PID file (best-effort - don't block shutdown)\n    const deleteResult = deletePidFile(pidFile, 'sigterm');\n    if (!deleteResult.success && deleteResult.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file during SIGTERM: ${deleteResult.error.message}`\n      );\n    }\n\n    process.exit(0);\n  });\n\n  // Create worker and execute job\n  const worker = new BackgroundWorker(\n    jobService,\n    services.store,\n    services.index,\n    services.lance,\n    services.embeddings\n  );\n\n  try {\n    await worker.executeJob(jobId);\n\n    // Clean up PID file on success (best-effort - don't change exit code)\n    const successCleanup = deletePidFile(pidFile, 'success');\n    if (!successCleanup.success && successCleanup.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file after success: ${successCleanup.error.message}`\n      );\n    }\n\n    console.log(`[${jobId}] Job completed successfully`);\n    process.exit(0);\n  } catch (error) {\n    // Job service already updated with failure status in BackgroundWorker\n    console.error(`[${jobId}] Job failed:`, error);\n\n    // Clean up PID file on failure (best-effort - exit code reflects job failure)\n    const failureCleanup = deletePidFile(pidFile, 'failure');\n    if (!failureCleanup.success && failureCleanup.error !== undefined) {\n      console.error(\n        `Warning: Could not remove PID file after failure: ${failureCleanup.error.message}`\n      );\n    }\n\n    process.exit(1);\n  }\n}\n\nmain().catch((error: unknown) => {\n  console.error('Fatal error in background worker:', error);\n  process.exit(1);\n});\n"],"mappings":";;;;;;;;;;;;;AAAA,SAAS,kBAAkB;AAkBpB,SAAS,uBACd,SACA,OACA,QAAgB,KACR;AACR,MAAI,UAAU,EAAG,QAAO;AACxB,SAAQ,UAAU,QAAS;AAC7B;AAEO,IAAM,mBAAN,MAAuB;AAAA,EAC5B,YACmB,YACA,cACA,cACA,YACA,iBACjB;AALiB;AACA;AACA;AACA;AACA;AAAA,EAChB;AAAA;AAAA;AAAA;AAAA,EAKH,MAAM,WAAW,OAA8B;AAC7C,UAAM,MAAM,KAAK,WAAW,OAAO,KAAK;AAExC,QAAI,CAAC,KAAK;AACR,YAAM,IAAI,MAAM,OAAO,KAAK,YAAY;AAAA,IAC1C;AAEA,QAAI;AAEF,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,YAAY,IAAI,IAAI;AAAA,QAC7B,UAAU;AAAA,QACV,SAAS,EAAE,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACjD,CAAC;AAGD,cAAQ,IAAI,MAAM;AAAA,QAChB,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF;AACE,gBAAM,IAAI,MAAM,qBAAqB,OAAO,IAAI,IAAI,CAAC,EAAE;AAAA,MAC3D;AAGA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,UAAU;AAAA,QACV,SAAS,GAAG,IAAI,IAAI;AAAA,QACpB,SAAS,EAAE,cAAa,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACnD,CAAC;AAAA,IACH,SAAS,OAAO;AAEd,YAAM,eAAwC;AAAA,QAC5C,cAAa,oBAAI,KAAK,GAAE,YAAY;AAAA,MACtC;AACA,UAAI,iBAAiB,SAAS,MAAM,UAAU,QAAW;AACvD,qBAAa,OAAO,IAAI,MAAM;AAAA,MAChC,OAAO;AACL,qBAAa,OAAO,IAAI,OAAO,KAAK;AAAA,MACtC;AACA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,iBAAiB,QAAQ,MAAM,UAAU;AAAA,QAClD,SAAS;AAAA,MACX,CAAC;AACD,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAMA,SAAK,WAAW,UAAU,IAAI,IAAI;AAAA,MAChC,QAAQ;AAAA,MACR,SAAS;AAAA,MACT,UAAU;AAAA,IACZ,CAAC;AAGD,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,gBAAgB,uBAAuB,MAAM,SAAS,MAAM,OAAO,EAAE;AAC3E,cAAM,gBAAgB,KAAK;AAE3B,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA;AAAA,UACpC,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,cAAc,cAAc,OAAO,CAAC;AAC1E,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAGA,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAEA,cAAM,WAAW,uBAAuB,MAAM,SAAS,MAAM,KAAK;AAElE,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,QAAQ;AAAA;AAAA,UAC/B,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,SAAS,KAAK,kBAAkB,oBAAoB,UAAU,QAAQ,YAAY,IACxF,IAAI;AAEN,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AACA,QAAI,QAAQ,UAAa,OAAO,QAAQ,UAAU;AAChD,YAAM,IAAI,MAAM,4BAA4B;AAAA,IAC9C;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,OAAO,SAAS,OAAO;AACzB,YAAM,IAAI,MAAM,aAAa,OAAO,YAAY;AAAA,IAClD;AAEA,UAAM,mBAAmB,OAAO,aAAa,WAAW,WAAW;AACnE,UAAM,UAAU,IAAI,mBAAmB;AAGvC,YAAQ,GAAG,YAAY,CAAC,aAA4B;AAElD,YAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,UAAI,YAAY,WAAW,aAAa;AACtC;AAAA,MACF;AAGA,YAAM,gBAAiB,SAAS,eAAe,mBAAoB;AAEnE,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SACE,SAAS,WACT,iBAAiB,OAAO,SAAS,YAAY,CAAC,IAAI,OAAO,gBAAgB,CAAC;AAAA,QAC5E,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA,QACpC,SAAS,EAAE,cAAc,SAAS,aAAa;AAAA,MACjD,CAAC;AAAA,IACH,CAAC;AAED,QAAI;AACF,YAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AACzC,YAAM,OAAmB,CAAC;AAG1B,YAAM,eAMF;AAAA,QACF,UAAU;AAAA,QACV,QAAQ,UAAU;AAAA,QAClB,aAAa,eAAe;AAAA;AAAA,MAC9B;AACA,UAAI,qBAAqB,QAAW;AAClC,qBAAa,mBAAmB;AAAA,MAClC;AACA,UAAI,uBAAuB,QAAW;AACpC,qBAAa,qBAAqB;AAAA,MACpC;AAGA,uBAAiB,UAAU,QAAQ,MAAM,KAAK,YAAY,GAAG;AAE3D,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,iBAAiB,OAAO,aAAa,OAAO;AAClD,cAAM,SAAS,MAAM,KAAK,gBAAgB,MAAM,cAAc;AAE9D,aAAK,KAAK;AAAA,UACR,IAAI,iBAAiB,GAAG,MAAM,EAAE,IAAI,WAAW,KAAK,EAAE,OAAO,OAAO,GAAG,EAAE,OAAO,KAAK,CAAC,EAAE;AAAA,UACxF,SAAS;AAAA,UACT;AAAA,UACA,UAAU;AAAA,YACR,MAAM;AAAA,YACN,SAAS,MAAM;AAAA,YACf,KAAK,OAAO;AAAA,YACZ,OAAO,OAAO;AAAA,YACd,WAAW,OAAO,cAAc;AAAA,YAChC,OAAO,OAAO;AAAA,YACd,WAAW,oBAAI,KAAK;AAAA,UACtB;AAAA,QACF,CAAC;AAAA,MACH;AAGA,UAAI,KAAK,SAAS,GAAG;AACnB,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS;AAAA,UACT,UAAU;AAAA,QACZ,CAAC;AAED,cAAM,KAAK,WAAW,aAAa,MAAM,IAAI,IAAI;AAEjD,cAAM,KAAK,WAAW,eAAe,MAAM,EAAE;AAAA,MAC/C;AAEA,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SAAS,uBAAuB,OAAO,KAAK,MAAM,CAAC;AAAA,QACnD,UAAU;AAAA,QACV,SAAS,EAAE,cAAc,KAAK,OAAO;AAAA,MACvC,CAAC;AAAA,IACH,UAAE;AACA,YAAM,QAAQ,KAAK;AAAA,IACrB;AAAA,EACF;AACF;;;ACvTA,OAAO,QAAQ;AACf,OAAO,UAAU;AA4BV,SAAS,aAAa,SAAiB,KAAmB;AAC/D,MAAI;AACF,OAAG,cAAc,SAAS,IAAI,SAAS,GAAG,OAAO;AAAA,EACnD,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,IAAI;AAAA,MACR,sCAAsC,OAAO,+DAExB,OAAO;AAAA,IAC9B;AAAA,EACF;AACF;AAeO,SAAS,cAAc,SAAiB,UAA+C;AAC5F,MAAI;AACF,OAAG,WAAW,OAAO;AACrB,WAAO,EAAE,SAAS,KAAK;AAAA,EACzB,SAAS,OAAO;AAEd,QAAI,iBAAiB,SAAS,UAAU,SAAS,MAAM,SAAS,UAAU;AACxE,aAAO,EAAE,SAAS,KAAK;AAAA,IACzB;AAEA,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;AAAA,IACjE;AAAA,EACF;AACF;AASO,SAAS,iBAAiB,SAAiB,OAAuB;AACvE,SAAO,KAAK,KAAK,SAAS,GAAG,KAAK,MAAM;AAC1C;;;ACnEA,eAAe,OAAsB;AACnC,QAAM,QAAQ,QAAQ,KAAK,CAAC;AAC5B,QAAM,UAAU,QAAQ,IAAI,iBAAiB;AAE7C,MAAI,UAAU,UAAa,UAAU,IAAI;AACvC,YAAQ,MAAM,wBAAwB;AACtC,YAAQ,MAAM,uCAAuC;AACrD,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,QAAM,aAAa,IAAI,WAAW,OAAO;AACzC,QAAM,WAAW,MAAM,eAAe,QAAW,OAAO;AAGxD,QAAM,UAAU;AAAA,IACd,WAAW,SAAS;AAAA;AAAA,IACpB;AAAA,EACF;AAEA,MAAI;AACF,iBAAa,SAAS,QAAQ,GAAG;AAAA,EACnC,SAAS,OAAO;AAEd,YAAQ,MAAM,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AACpE,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,UAAQ,GAAG,WAAW,MAAM;AAC1B,YAAQ,IAAI,IAAI,KAAK,uCAAuC;AAC5D,eAAW,UAAU,OAAO;AAAA,MAC1B,QAAQ;AAAA,MACR,SAAS;AAAA,IACX,CAAC;AAGD,UAAM,eAAe,cAAc,SAAS,SAAS;AACrD,QAAI,CAAC,aAAa,WAAW,aAAa,UAAU,QAAW;AAC7D,cAAQ;AAAA,QACN,sDAAsD,aAAa,MAAM,OAAO;AAAA,MAClF;AAAA,IACF;AAEA,YAAQ,KAAK,CAAC;AAAA,EAChB,CAAC;AAGD,QAAM,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,EACX;AAEA,MAAI;AACF,UAAM,OAAO,WAAW,KAAK;AAG7B,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,cAAQ;AAAA,QACN,qDAAqD,eAAe,MAAM,OAAO;AAAA,MACnF;AAAA,IACF;AAEA,YAAQ,IAAI,IAAI,KAAK,8BAA8B;AACnD,YAAQ,KAAK,CAAC;AAAA,EAChB,SAAS,OAAO;AAEd,YAAQ,MAAM,IAAI,KAAK,iBAAiB,KAAK;AAG7C,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,cAAQ;AAAA,QACN,qDAAqD,eAAe,MAAM,OAAO;AAAA,MACnF;AAAA,IACF;AAEA,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF;AAEA,KAAK,EAAE,MAAM,CAAC,UAAmB;AAC/B,UAAQ,MAAM,qCAAqC,KAAK;AACxD,UAAQ,KAAK,CAAC;AAChB,CAAC;","names":[]}

package/hooks/check-dependencies.sh

@@ -13,6 +13,29 @@ GREEN='\033[0;32m'
 YELLOW='\033[1;33m'
 NC='\033[0m' # No Color
 
+# =====================
+# Helper Functions
+# =====================
+
+# Install Playwright browser (called after crawl4ai is confirmed installed)
+install_playwright_browser() {
+    # Check if Playwright Chromium is already installed by testing if browser can be launched
+    if python3 -c "from playwright.sync_api import sync_playwright; p = sync_playwright().start(); b = p.chromium.launch(); b.close(); p.stop()" 2>/dev/null; then
+        echo -e "${GREEN}[bluera-knowledge] Playwright Chromium ready ✓${NC}"
+        return 0
+    fi
+
+    echo -e "${YELLOW}[bluera-knowledge] Installing Playwright browser (one-time setup)...${NC}"
+    if python3 -m playwright install chromium 2>/dev/null; then
+        echo -e "${GREEN}[bluera-knowledge] Playwright Chromium installed ✓${NC}"
+        return 0
+    else
+        echo -e "${YELLOW}[bluera-knowledge] Playwright browser install failed.${NC}"
+        echo -e "${YELLOW}Manual fix: python3 -m playwright install chromium${NC}"
+        return 1
+    fi
+}
+
 # =====================
 # Node.js Dependencies
 # =====================
@@ -60,6 +83,8 @@ if python3 -c "import crawl4ai" 2>/dev/null; then
     # Already installed - get version
     crawl4ai_version=$(python3 -c "import crawl4ai; print(crawl4ai.__version__)" 2>/dev/null || echo "unknown")
     echo -e "${GREEN}[bluera-knowledge] crawl4ai ${crawl4ai_version} is installed ✓${NC}"
+    # Ensure Playwright browser is installed for headless crawling
+    install_playwright_browser
     exit 0
 fi
 
@@ -87,12 +112,16 @@ if command -v pip3 &> /dev/null || command -v pip &> /dev/null; then
         echo -e "${GREEN}[bluera-knowledge] Successfully installed crawl4ai ✓${NC}"
         crawl4ai_version=$(python3 -c "import crawl4ai; print(crawl4ai.__version__)" 2>/dev/null || echo "installed")
         echo -e "${GREEN}[bluera-knowledge] crawl4ai ${crawl4ai_version} ready${NC}"
+        # Install Playwright browser for headless crawling
+        install_playwright_browser
     else
         # Fallback: try without --break-system-packages for older Python
         if $PIP_CMD install --quiet --user crawl4ai 2>/dev/null; then
             echo -e "${GREEN}[bluera-knowledge] Successfully installed crawl4ai ✓${NC}"
             crawl4ai_version=$(python3 -c "import crawl4ai; print(crawl4ai.__version__)" 2>/dev/null || echo "installed")
             echo -e "${GREEN}[bluera-knowledge] crawl4ai ${crawl4ai_version} ready${NC}"
+            # Install Playwright browser for headless crawling
+            install_playwright_browser
         else
             echo -e "${RED}[bluera-knowledge] Auto-installation failed${NC}"
             echo ""

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "bluera-knowledge",
-  "version": "0.9.42",
+  "version": "0.10.0",
   "description": "CLI tool for managing knowledge stores with semantic search",
   "type": "module",
   "bin": {

package/python/crawl_worker.py

@@ -29,10 +29,15 @@ async def fetch_headless(url: str):
         if not result.success:
             raise Exception(f"Crawl failed: {result.error_message}")
 
+        # Combine internal and external links - let TypeScript filter by domain
+        all_links = []
+        if isinstance(result.links, dict):
+            all_links = result.links.get("internal", []) + result.links.get("external", [])
+
         return {
             "html": result.html or '',
             "markdown": result.markdown or result.cleaned_html or '',
-            "links": result.links.get("internal", []) if isinstance(result.links, dict) else []
+            "links": all_links
         }
 
 def is_exported(node: ast.AST) -> bool:

package/src/cli/commands/crawl.test.ts

@@ -132,7 +132,7 @@ describe('crawl command execution', () => {
       expect(mockCrawler.crawl).toHaveBeenCalledWith('https://example.com', {
         crawlInstruction: 'all documentation pages',
         maxPages: 50,
-        useHeadless: false,
+        useHeadless: true,
       });
       expect(mockServices.embeddings.embed).toHaveBeenCalledTimes(2);
       expect(mockServices.lance.addDocuments).toHaveBeenCalledWith(
@@ -162,6 +162,46 @@ describe('crawl command execution', () => {
       expect(processExitSpy).not.toHaveBeenCalled();
     });
 
+    it('uses axios-only mode when --fast flag is provided', async () => {
+      const mockStore: WebStore = {
+        id: createStoreId('store-1'),
+        name: 'test-store',
+        type: 'web',
+        url: 'https://example.com',
+        depth: 2,
+        createdAt: new Date(),
+        updatedAt: new Date(),
+      };
+
+      mockServices.store.getByIdOrName.mockResolvedValue(mockStore);
+      mockServices.lance.initialize.mockResolvedValue(undefined);
+      mockServices.embeddings.embed.mockResolvedValue([0.1, 0.2, 0.3]);
+      mockServices.lance.addDocuments.mockResolvedValue(undefined);
+
+      mockCrawler.crawl.mockReturnValue(
+        (async function* () {
+          yield {
+            url: 'https://example.com/page1',
+            title: 'Page 1',
+            markdown: '# Page 1',
+            depth: 0,
+          };
+        })()
+      );
+
+      const command = createCrawlCommand(getOptions);
+      command.parseOptions(['--fast', '--max-pages', '25']);
+      const actionHandler = command._actionHandler;
+
+      await actionHandler(['https://example.com', 'test-store']);
+
+      expect(mockCrawler.crawl).toHaveBeenCalledWith('https://example.com', {
+        maxPages: 25,
+        useHeadless: false, // --fast disables headless
+      });
+      expect(mockCrawler.stop).toHaveBeenCalled();
+    });
+
     it('successfully crawls in simple mode', async () => {
       const mockStore: WebStore = {
         id: createStoreId('store-1'),
@@ -198,7 +238,7 @@ describe('crawl command execution', () => {
       expect(mockCrawler.crawl).toHaveBeenCalledWith('https://example.com', {
         maxPages: 25,
         simple: true,
-        useHeadless: false,
+        useHeadless: true,
       });
       expect(mockCrawler.stop).toHaveBeenCalled();
     });
@@ -295,7 +335,7 @@ describe('crawl command execution', () => {
         crawlInstruction: 'all Getting Started pages',
         extractInstruction: 'code examples',
         maxPages: 100,
-        useHeadless: false,
+        useHeadless: true,
       });
     });
   });

package/src/cli/commands/crawl.ts

@@ -25,7 +25,7 @@ export function createCrawlCommand(getOptions: () => GlobalOptions): Command {
     )
     .option('--simple', 'Use simple BFS mode instead of intelligent crawling')
     .option('--max-pages <number>', 'Maximum number of pages to crawl', '50')
-    .option('--headless', 'Use headless browser for JavaScript-rendered sites')
+    .option('--fast', 'Use fast axios-only mode (may fail on JavaScript-heavy sites)')
     .action(
       async (
         url: string,
@@ -35,7 +35,7 @@ export function createCrawlCommand(getOptions: () => GlobalOptions): Command {
           extract?: string;
           simple?: boolean;
           maxPages?: string;
-          headless?: boolean;
+          fast?: boolean;
         }
       ) => {
         const globalOpts = getOptions();
@@ -125,7 +125,7 @@ export function createCrawlCommand(getOptions: () => GlobalOptions): Command {
             ...(cmdOptions.extract !== undefined && { extractInstruction: cmdOptions.extract }),
             maxPages,
             ...(cmdOptions.simple !== undefined && { simple: cmdOptions.simple }),
-            useHeadless: cmdOptions.headless ?? false,
+            useHeadless: !(cmdOptions.fast ?? false), // Default true (headless), --fast disables
           })) {
             // Use extracted content if available, otherwise markdown
             const contentToProcess = result.extracted ?? result.markdown;

package/src/crawl/claude-client.test.ts

@@ -81,7 +81,11 @@ describe('ClaudeClient', () => {
 
   describe('determineCrawlUrls', () => {
     it('should successfully parse valid crawl strategy response', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all docs');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all docs'
+      );
 
       // Simulate successful response
       setTimeout(() => {
@@ -103,7 +107,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should call spawn with correct arguments for determineCrawlUrls', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all docs');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all docs'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -136,7 +144,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should write prompt to stdin', async () => {
-      const promise = client.determineCrawlUrls('<html><body>Test</body></html>', 'Find tutorials');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html><body>Test</body></html>',
+        'Find tutorials'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -160,7 +172,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when response has no urls array', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -178,7 +194,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when response has empty urls array', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -197,7 +217,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when response has no reasoning', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -215,7 +239,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when urls contains non-string values', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -234,7 +262,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when response is not valid JSON', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('Not valid JSON'));
@@ -245,7 +277,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should reject when response is null', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('null'));
@@ -257,7 +293,7 @@ describe('ClaudeClient', () => {
 
     it('should truncate HTML longer than 50000 characters', async () => {
       const longHtml = '<html>' + 'a'.repeat(60000) + '</html>';
-      const promise = client.determineCrawlUrls(longHtml, 'Find all');
+      const promise = client.determineCrawlUrls('https://example.com', longHtml, 'Find all');
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -281,7 +317,7 @@ describe('ClaudeClient', () => {
 
     it('should not truncate HTML shorter than 50000 characters', async () => {
       const shortHtml = '<html><body>Short content</body></html>';
-      const promise = client.determineCrawlUrls(shortHtml, 'Find all');
+      const promise = client.determineCrawlUrls('https://example.com', shortHtml, 'Find all');
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -302,6 +338,33 @@ describe('ClaudeClient', () => {
       expect(writtenPrompt).toContain(shortHtml);
       expect(writtenPrompt).not.toContain('[... HTML truncated ...]');
     });
+
+    it('should include seedUrl in prompt for relative URL resolution', async () => {
+      const promise = client.determineCrawlUrls(
+        'https://code.claude.com/docs',
+        '<html><a href="/docs/en/hooks">Hooks</a></html>',
+        'Find all docs'
+      );
+
+      setTimeout(() => {
+        mockProcess.stdout.emit(
+          'data',
+          Buffer.from(
+            JSON.stringify({
+              urls: ['https://code.claude.com/docs/en/hooks'],
+              reasoning: 'Found hooks documentation',
+            })
+          )
+        );
+        mockProcess.emit('close', 0);
+      }, 10);
+
+      await promise;
+
+      const writtenPrompt = vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string;
+      expect(writtenPrompt).toContain('Base URL: https://code.claude.com/docs');
+      expect(writtenPrompt).toContain('resolve them against the Base URL');
+    });
   });
 
   describe('extractContent', () => {
@@ -386,7 +449,11 @@ describe('ClaudeClient', () => {
 
   describe('Subprocess Management', () => {
     it('should handle process spawn errors', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.emit('error', new Error('spawn ENOENT'));
@@ -410,7 +477,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should collect stderr data', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stderr.emit('data', Buffer.from('Error message 1\n'));
@@ -466,14 +537,22 @@ describe('ClaudeClient', () => {
 
   describe('Timeout Handling', () => {
     it('should timeout after configured timeout period', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       // Don't emit close event - let it timeout
       await expect(promise).rejects.toThrow('timed out after 100ms');
     });
 
     it('should kill process on timeout', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       await expect(promise).rejects.toThrow('timed out');
       expect(mockProcess.kill).toHaveBeenCalledWith('SIGTERM');
@@ -540,7 +619,11 @@ describe('ClaudeClient', () => {
 
   describe('JSON Parsing', () => {
     it('should handle malformed JSON in response', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('{ invalid json }'));
@@ -551,7 +634,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should handle incomplete JSON in response', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('{"urls": ["https://example.com"'));
@@ -562,7 +649,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should handle JSON with extra whitespace', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -582,7 +673,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should handle JSON arrays as invalid for determineCrawlUrls', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('[]'));
@@ -593,7 +688,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should handle JSON primitives as invalid for determineCrawlUrls', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit('data', Buffer.from('"string response"'));
@@ -606,7 +705,11 @@ describe('ClaudeClient', () => {
 
   describe('Response Validation', () => {
     it('should validate urls is an array', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -625,7 +728,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should validate reasoning is a string', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -644,7 +751,11 @@ describe('ClaudeClient', () => {
     });
 
     it('should accept valid response with multiple URLs', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.stdout.emit(
@@ -671,7 +782,11 @@ describe('ClaudeClient', () => {
 
   describe('Error Messages', () => {
     it('should wrap errors with context for determineCrawlUrls', async () => {
-      const promise = client.determineCrawlUrls('<html>test</html>', 'Find all');
+      const promise = client.determineCrawlUrls(
+        'https://example.com',
+        '<html>test</html>',
+        'Find all'
+      );
 
       setTimeout(() => {
         mockProcess.emit('close', 1);