Sample cases

Published sample cases and proof bundles

Start with the AI Agent Action Proof Run sample. It shows how a third party inspects the action boundary, authority map, evidence manifest, receipt, verifier result, challenge path, and digest manifest after an agent acts.

AI Agent Action Proof Run

A public standalone bundle for one agent-assisted code or configuration change. Use it to inspect the receipt shape and verifier path before submitting your own workflow.

Artifact set

  • ACTION_BOUNDARY.json
  • AUTHORITY_MAP.json
  • EVIDENCE_MANIFEST.json
  • RECEIPT.json
  • VERIFY_RESULT.json
  • CHALLENGE_PATH.md
  • MANIFEST.sha256

What these pages are for

Show one bounded workflow class

Each page keeps authority, execution, evidence, verification, and challenge language separate.

Name the limits

Samples are explanatory. They do not imply live customer proof, production deployment, legal compliance, or complete AI governance coverage.

Supporting surfaces

Additional named sample cases

Privileged access grant review

A named sample case for a time-bounded administrative access request, focused on authority binding, provisioning execution, resulting entitlement evidence, and replayability outside the source systems.

  • approval to execution continuity
  • provisioning evidence
  • target-side entitlement confirmation

Open named sample case

Approval-gated containment review

A named sample case for a containment action that must not execute before recorded approval, focused on gate enforcement, target-state evidence, and portable replay after the event.

  • pre-execution gate enforcement
  • target-side containment state
  • portable replay path

Open named sample case