See what the agent was allowed to do.
Every checked action leaves a proof of what was checked, which policy applied, and what was allowed or refused.
Definition
An agent audit trail is evidence another system can check: what was verified, which policy applied, and what actually happened.
What the trail records
The policy is on record
The delegation proof shows the allowed actions, spend caps, and expiry.
The decision is on record
Each checked action leaves a proof with the agent, its owner, and the result.
The record is verifiable
Another system rechecks the proof and sees whether it still holds.
Read the record after the action
receipt.ts
When to use it
Use it when
- Regulated workflows need evidence of who approved what.
- Payments or payouts need a receipt tied to the policy that allowed them.
- Incident review needs what the agent was permitted to do, not just what it did.
Not when
- Read-only browsing that touches nothing privileged.
- Tracing you already own and trust inside one closed system.
Evidence boundaries
- The record shows what was checked and what was allowed. Private payloads stay private.
- Records carry status and expiry. Old evidence does not pass as current.
- Revoked proof is visible as revoked. The trail never hides a refusal.

Keep evidence of every decision.
Checked actions leave proof with the agent, its owner, and the result. Another system can verify it later.