Clawb Docs

Clawb is an agent control plane for production systems. When an agent asks your system to do something sensitive (refund money, deploy to prod, call a paid API), your backend uses Clawb’s control plane to verify identity and enforce a policy decision:

• allow → proceed
• challenge → require extra proof (human approval, 2FA, business-hours rule, telemetry threshold, etc.)
• deny → block

​

What Clawb controls at runtime

• Identity verification: verify signed agent requests (/v1/verify).
• Policy decisioning: evaluate action context and return allow | challenge | deny (/v1/check).
• Bounded credentials: mint short-lived credentials for follow-on provider calls.
• Audit + incident response: query/export evidence and use kill switch controls during incidents.

​

60-second integration path

1. Create and attest an agent passport (agent_id + key).
2. Verify inbound signed requests in your backend service or gateway.
3. Call /v1/check immediately before sensitive actions.
4. Enforce allow | challenge | deny in code.
5. Capture trace/audit signals and wire incident controls.

​

Decision semantics

​

Who this is for

Backend integrators, enterprise backend teams, and platform/security teams shipping agent-powered production systems.

​

Start here

• Quickstart
• End-to-end walkthrough
• Core concepts: agents
• Core concepts: policies
• Build your integration (/v1/check)
• Python SDK
• Operate in production
• API reference