Controlled Actions
Controlled Actions define what a registered AI agent may do automatically, what must be escalated, what is restricted, and what must be blocked before execution. This page now reads agents registered inside Agent Registry and creates Boundary Decision Packages when action checks run.
Select a registered agent to inspect its runtime action matrix.
If no registered agent exists, use Agent Registry first. Proof-derived agents can still be shown from existing evidence.
Agent Action Policy
Select an action to inspect its default decision, enforcement mode, fail behavior, required authority, proof receipt, and downstream sync.
| Controlled Action | Mode | Fail Behavior | Default Decision | Required Authority | Proof Receipt | Inspect |
|---|---|---|---|---|---|---|
| Loading controlled actions... | ||||||