Guides
Self-hosting overview
Understand the three-part deployment model and where ClawForge data lives.
Three deployment surfaces
ClawForge is organized around three moving pieces:
- runtime adapters that enforce policy close to the assistant
- a control plane API that stores policy and audit state
- an operator console for review, approvals, and incident response
Trust boundary summary
For self-hosted teams, the important question is simple:
What stays under your control?
- control plane API
- admin console
- PostgreSQL state
- audit storage and retention policy
- identity provider configuration
What to verify during evaluation
- Where your audit data is stored
- How policies are versioned and rolled out
- How kill-switch changes propagate
- Which identity path you want to use at launch