Audit trail

Show the system saying no, then show why.

The audit page gives judges a clean way to verify that Smart Spend Sentinel is a policy engine with consequences.

Decision timeline

Approved x402 market-intel endpoint

2026-06-08 10:24

Risk 24/65. Weekly cap 35 USD. Permission expires in 72 hours.

Blocked GPU compute burst

2026-06-08 10:31

Requested 92.00 USD exceeds policy cap 40.00 USD. Compute category requires strict approval.

Approved Venice AI inference refill

2026-06-08 10:38

Risk 18/45. Strict profile allowed short AI burst inside 20 USD cap.

Budget cap

Every request is compared with the user policy before package assembly.

Category risk

Compute, storage, AI inference, data APIs, and unknown categories score differently.

Expiry window

Permissions carry a concrete validity window instead of permanent delegation.

Receipt proof

Decision, hash, and relayer plan are visible for post-run inspection.