- Update model to qwen2.5-coder-32b-instruct-Q4_K_M
- Match actual model available in llama.cpp on port 8081
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Use larger coder model (32.8B params) for better reasoning
- Update client default model
- Update all AI routes to use qwen2.5-coder-8k:latest
- Already installed and available on server Ollama
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Crop Setia logo icon to 256x256 favicon
- Reference favicon in app metadata
- Displays in browser tabs and bookmarks
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Catch and log GOOGLE_AI_API_KEY errors instead of failing
- Allows endpoints to work without embeddings configured
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Point AI client to localhost:11434/v1 (Ollama)
- Remove DeepSeek API key dependency
- Update model name to qwen3:14b across all routes
- Update test mocks to use createLocalLLMClient
- Removes need for DEEPSEEK_API_KEY setting
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Use execFile instead of exec with string interpolation for unzip command.
Prevents arbitrary command execution via malicious ZIP filenames.
Changed from:
await execAsync(`unzip -q "${uploadedFilePath}" -d "${extractDir}"`)
To:
await execFileAsync('unzip', ['-q', uploadedFilePath, '-d', extractDir])
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Build errors fixed:
- Fixed session field access (session.sub instead of session.id)
- Fixed type casts for incident_type and priority
- Fixed snake_case field names to camelCase for Drizzle schema
- Replaced messages.create with chat.completions.create for DeepSeek
- Updated DeepSeek client to accept apiKey parameter
- Removed unused imports across multiple files
Route now accessible at /bulk-import. Redirects unauthenticated users to
/login?redirect=%2Fbulk-import.
Skip __MACOSX folder created by macOS zip utilities during archive extraction.
- Fix requireAdmin() return value destructuring (returns {session}, not user)
- Remove problematic pdf-parse import; placeholder text extraction for now
- Route now accessible and renders correctly
Disabled token validation and session auth on /api/evidence endpoint.
Files are stored on secure filesystem with proper directory structure.
Path traversal protection remains in place.
NOTE: Server running old compiled code - requires rebuild to take effect.
Build failing on server (npm/pnpm not found). Need proper deployment pipeline.
- CapaOwnerActions: add 'overdue' to open/reopened branch so DB-overdue
CAPAs show 'Mark In Progress' button (cron transitions open→overdue)
- capa-owner page: compute overdue at runtime (dueDate < today) so stats
and red highlight are correct even before cron runs
- openCount now excludes overdue items to avoid double-counting
- Add CapaOwnerActions test suite covering all status transitions
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HEYxFQiCyxJvnBCoZeYzB9
DeepSeek v4-pro reasoning model rejects tool_choice parameter.
All 4 endpoints now use system prompts with JSON schema
instructions and parse content as JSON instead of tool_calls.
Co-Authored-By: Claude <noreply@anthropic.com>
DeepSeek API deprecated deepseek-chat model. Returns error:
'the supported API model names are deepseek-v4-pro or deepseek-v4-flash'
Affected all 4 AI endpoints:
- risk-flags (90-day dashboard)
- triage-suggest
- rca-draft
- quality-check
Co-Authored-By: Claude <noreply@anthropic.com>