One dashboard to govern all your AI assistants.
ClawForge is the admin layer for OpenClaw. It gives organizations centralized control over authentication, tool policies, skill governance, audit trails, and a remote kill switch — across every OpenClaw instance in the fleet.
- Centralized Policy Enforcement — Define tool allow/deny lists and enforcement profiles from one place
- Skill Governance — Employees submit skills for admin review; only approved skills run in production
- Audit Trail — Every tool call, session, and LLM interaction is logged and queryable
- Kill Switch — Instantly disable all AI tool access across the entire org in an emergency
- SSO / OIDC — Okta, Auth0, Entra ID, or any OIDC provider, alongside email/password auth
- Enrollment Tokens — Generate invite tokens to onboard new users without SSO
- Heartbeat Monitoring — Know which instances are online and what policy version they're running
/clawforge-status— Users can see their auth state, org, policy version, and kill switch status
git clone https://github.com/ClawForgeAI/clawforge.git
cd clawforge
docker compose up --buildOnce running:
- Admin Console — localhost:4200
- API — localhost:4100
- Login —
admin@clawforge.local/clawforge
For manual setup, SSO configuration, and connecting an OpenClaw gateway, see the Setup Guide.
| Guide | Description |
|---|---|
| Setup Guide | Docker, manual setup, SSO, connecting OpenClaw |
| Architecture & How It Works | Package structure, core concepts, data flows, database schema |
| API Reference | Every endpoint with request/response examples |
| Configuration | Plugin config, server env vars, admin console env vars |
| Roadmap | What's next for ClawForge |
Contributions are welcome! Please open an issue or submit a pull request.
MIT