Crew Runner

CrewAI execution engine for Alteriom Command Center

Crew Runner is a standalone microservice that executes CrewAI workflows on behalf of the Alteriom Command Center backend. It provides a simple HTTP API for submitting crew execution requests and retrieving results.

---

Features

• ✅ FastAPI HTTP API - Simple REST interface for crew execution
• ✅ LiteLLM Integration - Unified API for multiple LLM providers (OpenAI, Anthropic, etc.)
• ✅ Worker Registration - Auto-registers with backend on startup
• ✅ Health Monitoring - Built-in health checks and heartbeat system
• ✅ Version Management - Semantic versioning from VERSION file
• ✅ Docker Support - Production-ready containerized deployment
• ✅ Zero Zombie Processes - Pure Python/FastAPI (no subprocess spawning)

---

Quick Start

Local Development

# Install dependencies
pip install -r requirements.txt

# Set environment variables
export LITELLM_API_KEY=sk-...
export BACKEND_URL=http://localhost:3007
export OLLAMA_BASE_URL=http://localhost:11434

# Run server
python src/main.py

Server runs on http://localhost:8081

Docker

# Build image
docker build -t crew-runner:latest .

# Run container
docker run -d \
  --name crew-runner \
  -p 8081:8081 \
  -e LITELLM_API_KEY=sk-... \
  -e BACKEND_URL=http://backend:3007 \
  -e OLLAMA_BASE_URL=http://ollama:11434 \
  crew-runner:latest

Docker Compose

version: '3.8'
services:
  crew-runner:
    image: ghcr.io/alteriom/crew-runner:1.0.0
    ports:
      - "127.0.0.1:8081:8081"
    environment:
      - LITELLM_API_KEY=${LITELLM_API_KEY}
      - BACKEND_URL=http://backend:3007
      - OLLAMA_BASE_URL=http://ollama:11434
      - MAX_SESSIONS=2
      - WORKER_HEARTBEAT_INTERVAL=60
    restart: unless-stopped
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8081/health"]
      interval: 30s
      timeout: 10s
      retries: 3

---

API Endpoints

GET /health

Health check endpoint

Response:

{
  "status": "ok",
  "version": "1.0.0",
  "active_sessions": 0,
  "max_sessions": 2,
  "at_capacity": false
}

GET /version

Version information

Response:

{
  "version": "1.0.0",
  "service": "crew-runner"
}

POST /execute

Execute a CrewAI workflow

Request:

{
  "prompt": "Analyze this data...",
  "system_context": "You are a data analyst",
  "execution_id": "exec-123",
  "tenant_id": "tenant-456",
  "inputs": {},
  "timeout_seconds": 300
}

Response:

{
  "result": "Analysis complete...",
  "execution_id": "exec-123",
  "status": "completed"
}

---

Environment Variables

| Variable | Required | Default | Description | |----------|----------|---------|-------------| | LITELLM_API_KEY | Yes | - | LiteLLM API key for LLM access | | BACKEND_URL | Yes | - | Command Center backend URL | | OLLAMA_BASE_URL | No | http://localhost:11434 | Ollama server URL | | MAX_SESSIONS | No | 2 | Maximum concurrent executions | | WORKER_HEARTBEAT_INTERVAL | No | 60 | Heartbeat interval (seconds) | | DEBUG | No | false | Enable debug logging |

---

Integration with Command Center

Crew Runner integrates with the Alteriom Command Center backend:

1. Worker Registration: On startup, crew-runner registers itself with the backend
2. Execution Requests: Backend sends execution requests via POST /execute
3. Heartbeat: crew-runner sends periodic heartbeats to backend
4. Results: Execution results are returned in the response

Backend Compatibility

| crew-runner | Command Center Backend | |-------------|------------------------| | 1.0.x | >= 1.5.0 |

---

Development

Running Tests

# Set test environment
export CREW_RUNNER_URL=http://localhost:8081
export BACKEND_URL=http://localhost:3007

# Run test suite
./scripts/test-config.sh

Project Structure

crew-runner/
├── src/
│   ├── main.py                    # FastAPI application
│   └── llm_output_normalizer.py   # LLM output processing
├── tests/                         # Unit tests
├── scripts/
│   └── test-config.sh            # Integration test script
├── .github/workflows/
│   └── ci.yml                    # GitHub Actions CI/CD
├── Dockerfile                     # Production container
├── docker-compose.yml            # Local dev environment
├── requirements.txt              # Python dependencies
├── VERSION                       # Semantic version
├── SECURITY.md                   # Security policy
└── README.md                     # This file

Making Changes

1. Fork the repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Make your changes
4. Run tests locally
5. Commit your changes (git commit -am 'Add new feature')
6. Push to the branch (git push origin feature/my-feature)
7. Create a Pull Request

---

Deployment

GitHub Container Registry

Images are automatically built and published to GHCR on every push to main:

# Pull latest image
docker pull ghcr.io/alteriom/crew-runner:latest

# Pull specific version
docker pull ghcr.io/alteriom/crew-runner:1.0.0

Production Deployment

1. Pull the desired version from GHCR
2. Update docker-compose.prod.yml with the version tag
3. Deploy:

   docker compose -f docker-compose.prod.yml up -d crew-runner
   

Version Pinning

Always pin versions in production:

services:
  crew-runner:
    image: ghcr.io/alteriom/crew-runner:1.0.0  # ✅ Pinned
    # NOT: ghcr.io/alteriom/crew-runner:latest  # ❌ Unpredictable

---

Security

Current Status (v1.0.0)

⚠️ No authentication implemented

Crew-runner v1.0.0 should only be deployed:

• Behind a firewall
• On internal networks only
• With access restricted to trusted backend services

Planned (v1.1.0)

• ✅ API key authentication
• ✅ Rate limiting
• ✅ Input sanitization
• ✅ Request signing (HMAC)

See SECURITY.md for full security policy.

---

Monitoring

Health Checks

# Check service health
curl http://localhost:8081/health

# Check version
curl http://localhost:8081/version

Logs

# Docker logs
docker logs crew-runner --tail 100 --follow

# Check for errors
docker logs crew-runner 2>&1 | grep -i error

Metrics

Crew-runner logs execution metrics:

• Execution ID
• Duration
• Status (success/failure)
• Error messages (if any)

---

Troubleshooting

Connection Refused

Problem: Connection refused on port 8081
Solution: Check if crew-runner is running and ports are correctly mapped

docker ps | grep crew-runner
netstat -tuln | grep 8081

Backend Registration Failed

Problem: crew-runner can't register with backend
Solution: Verify BACKEND_URL is correct and backend is reachable

curl $BACKEND_URL/health

Execution Timeouts

Problem: Crews timeout or hang
Solution: Increase timeout_seconds in execution requests or check LiteLLM API key

---

License

MIT License - see LICENSE for details

---

Contributing

Contributions are welcome! Please read our Contributing Guide for details on our code of conduct and the process for submitting pull requests.

---

Support

• Issues: GitHub Issues
• Documentation: GitHub Wiki
• Email: [email protected]

---

Version: 1.0.0
Last Updated: April 14, 2026