Create mock backend to simulate AI responses for UI development:
Backend Implementation:
- tools/fake-backend/server.js: Express server with CORS
- POST /api/analyze: Accepts log, returns mock AI analysis with 1.5s delay
- GET /health: Health check endpoint
- Pattern matching for different log types (PostgreSQL, Nginx, Node.js, Disk)
- Error handling: 400 for empty payload, 500 for server errors
- Mock responses for common errors (OOM, 502, connection refused, disk full)
Container Setup:
- Dockerfile: Node.js 20 Alpine container
- docker-compose.yml: Added fake-backend service on port 3000
- Health checks for both frontend and backend services
- Environment variable VITE_API_URL for frontend
Frontend Integration:
- InteractiveDemo.tsx: Replaced static data with real fetch() calls
- API_URL configurable via env var (default: http://localhost:3000)
- Error handling with user-friendly messages
- Shows backend URL in demo section
- Maintains loading states and UI feedback
Documentation:
- docs/tools_fake_backend.md: Complete usage guide
- README.md: Updated with tools/fake-backend structure and usage
Development Workflow:
1. docker compose up -d (starts both frontend and backend)
2. Frontend calls http://fake-backend:3000/api/analyze
3. Backend returns realistic mock responses
4. No OpenRouter API costs during development
Safety First:
- No real API calls during development
- Isolated mock logic in dedicated tool
- Easy switch to real backend by changing URL
- CORS enabled only for development
Refs: Sprint 4 preparation, API development workflow
