Free browser-local tool
Validate an agent skill permission manifest before trust is granted.
Paste a filled manifest JSON. This page checks the core structure and the safety controls that matter most: scoped authority, human approval for risky actions, memory promotion rules, receipts, exception handling, eval coverage, and review verdict.
Manifest JSON
Nothing is uploaded. Validation runs in your browser, so you can redact sensitive details before requesting a paid mini-report.