Built for Security and Flexibility
OpenPact gives you complete control over your AI assistant while keeping your data private and secure.
Security First
Principle of least privilege - your AI can only use tools you explicitly allow. Secrets never leak to the AI model.
75+ AI Providers
Powered by OpenCode, giving you access to Anthropic, OpenAI, Google, Ollama, and 75+ more LLM providers. Switch models without changing your setup.
Sandboxed Scripting
Extend capabilities with Starlark scripts. Python-like syntax, but safely sandboxed with no filesystem access.
Production Ready
Docker-native deployment with health checks, Prometheus metrics, structured logging, and rate limiting built in.
How It Works
OpenPact sits between you and your AI, ensuring security and control at every step.
Built for Everyone
Whether you're an individual, developer, or enterprise team, OpenPact adapts to your needs.
Privacy-Conscious Users
Keep your data local
- All data stays on your machine
- No telemetry or tracking
- Full control over what AI sees
- Encrypted communications
Developers
Build custom integrations with Starlark
- Python-like scripting syntax
- HTTP requests and JSON handling
- Safely sandboxed execution
- Easy to test and deploy
Teams
Audit and control what your AI can access
- Approve scripts before execution
- Role-based access control
- Complete audit trail
- Centralized secret management
Self-Hosters
Full control over your AI stack
- Docker-native deployment
- Prometheus metrics built-in
- Health checks and monitoring
- Use your own LLM provider
Get Started in Seconds
One command is all you need to run your own secure AI assistant.
docker run -d \
-p 8080:8080 \
-v openpact-workspace:/workspace \
-e DISCORD_TOKEN=your_token \
ghcr.io/open-pact/openpact:latest