AI Policies

Rule trees that govern every agent action — block, require approval, warn, or log.

Policies
Enabled
Blocking
Critical / High
Violations (recent)
PolicySeverityOn violationFrameworks ViolationsUpdatedEnabled
loading…

Simulate & evaluate

POST a candidate action context to /api/policies/evaluate. Every enabled policy is evaluated; the strictest matching action wins.

Action context
Verdict
Run an evaluation to see the verdict and which policies matched.