An AI-powered coding assistant that can control your computer, write code, debug issues, and automate development tasks.
- AI Code Generation - Generate code in any language with natural language descriptions
- Intelligent Debugging - Analyze and fix bugs with AI assistance
- Code Refactoring - Improve code quality and performance automatically
- Test Generation - Create comprehensive test suites
- Documentation - Generate API docs, READMEs, and code comments
- Computer Control - AI can control desktop, run commands, and use IDEs
- Full Monitoring - Production-ready logging, metrics, and alerting
- Docker & Docker Compose
- Anthropic API Key (Get one here)
- 8GB+ RAM recommended
- 20GB+ free disk space
Edit .env file and add your Anthropic API key:
ANTHROPIC_API_KEY=sk-ant-api03-xxxxx..../scripts/start.sh- Main UI: http://localhost:8080
- API Docs: http://localhost:8000/docs
- VNC Desktop: http://localhost:6080/vnc.html
- Code Server: http://localhost:8443
- Grafana: http://localhost:3000 (admin/admin)
- Open http://localhost:8080
- Enter your coding task
- Select language and preferences
- Click "Generate Code"
- View results and download code
import requests
response = requests.post(
"http://localhost:8000/execute",
json={
"task": "Create a Python REST API for todo management",
"context": {"language": "python", "framework": "fastapi"}
}
)
print(response.json())curl -X POST "http://localhost:8000/execute" \
-H "Content-Type: application/json" \
-d '{"task": "Write a sorting algorithm in Python"}'See .env file for all configuration options:
ANTHROPIC_API_KEY- Your API key (required)LOG_LEVEL- Logging level (INFO, DEBUG, ERROR)VNC_PASSWORD- VNC access password- Alert configurations (Email, Slack, PagerDuty)
| Service | Port | Description |
|---|---|---|
| Streamlit | 8080 | Main web interface |
| FastAPI | 8000 | REST API |
| VNC | 5900 | VNC desktop access |
| noVNC | 6080 | Web-based VNC |
| Code Server | 8443 | VS Code in browser |
| Grafana | 3000 | Monitoring dashboards |
| Prometheus | 9091 | Metrics |
- Grafana: http://localhost:3000
- Username: admin
- Password: admin (change on first login)
- Prometheus: http://localhost:9091
- Jaeger: http://localhost:16686 (tracing)
- Request rate and latency
- Error rates by type
- Task completion times
- System resources (CPU, memory, disk)
- Active sessions
Configure alerts in .env:
- Email alerts via SMTP
- Slack webhooks
- PagerDuty integration
./scripts/start.sh./scripts/stop.shdocker-compose logs -f./scripts/update.sh./scripts/backup.shdocker exec -it computer-use-coding bashcomputer-use-coding-assistant/
ββ
βββ agent/ # AI agent logic
βββ api/ # FastAPI server
βββ monitoring/ # Monitoring tools
βββ docker/ # Docker configuration
βββ monitoring/ # Monitoring configs
β βββ grafana/ # Grafana dashboards
β βββ prometheus/ # Prometheus config
βββ projects/ # Project files
βββ logs/ # Application logs
βββ scripts/ # Utility scripts
βββ docker-compose.yml # Service orchestration
βββ .env # Configuration
βββ README.md # Documentation
- Verify key in
.envfile - Check key has sufficient credits
- Ensure key is active
- Modify ports in
docker-compose.yml - Check for conflicting services
# Check container status
docker-compose ps
# View logs
docker-compose logs coding-assistant
# Restart services
docker-compose restart- Allocate more Docker resources
- Check system requirements
- Monitor via Grafana dashboards
- Run in isolated Docker environment
- VNC password protected
- API key stored securely
- Automatic log redaction
- Network isolation
- Check logs:
docker-compose logs -f - API documentation: http://localhost:8000/docs
- Monitoring dashboard: http://localhost:3000
MIT License - See LICENSE file for details
Built with:
