agent-governance-audit

What problem does it solve?

This Skill determines whether an AI-assisted change actually followed the required governance process, instead of merely claiming that it did. It helps catch missing approvals, improper merge authority, weak validation, incomplete closeouts, and undocumented memory or policy updates.

Core Features & Use Cases

• Primary-evidence audit: Cross-check PR metadata, timeline events, commit history, and CI logs against the policy in force at the time of the change.
• Control-by-control verdicts: Produce PASS, FAIL, or UNVERIFIABLE results for classification, approvals, merge authority, validation, security review, closeout completeness, and memory/doc governance.
• Post-incident and autonomy checks: Use it after an incident, before expanding agent autonomy, or when verifying whether a closeout report matches the real record.
• Example: A team asks whether a merged PR respected human approval and auto-merge rules; this Skill traces who armed auto-merge, what checks ran, and whether the closeout matches the primary evidence.

Quick Start

Ask for a governance audit of the change or PR, and provide the PR number or commit range plus the policy baseline date if known.