What problem does it solve? Compliance evidence drafted by AI agents can contain dangling evidence pointers, pre-signed attestations, or fabricated results that slip into the ledger. This Skill adds a second review pass on compliance-evidence pull requests so the human executive signs off on verified, complete evidence instead of unvetted drafts. ## Core Features & Use Cases - Evidence Verification: Confirms that evidence pointers in compliance/evidence-ledger.yml resolve to real files and that the evidence doc contains substantive assessment content rather than empty templates. - Anti-Laundering Checks: Detects pre-signing attempts where a draft claims source: human-attested without a human edit, and flags too-good-to-be-true results with zero findings for extra executive scrutiny. - Advisory Status Reporting: Posts a compliance-check status (success/failure) with findings and guidance on what the executive should examine before signing, without blocking the merge. - Use Case: When a PR appends a new SOC 2 access-review artifact to the evidence ledger, this Skill checks that the collaborator table is real, the interval matches the issue, and findings are surfaced, then comments with anything the executive must verify before approval. ## Quick Start When a pull request touching compliance/evidence-ledger.yml opens, review the drafter's evidence for missing or wrong content and post an advisory compliance-check status with your findings.
