What problem does it solve? AI policies drift out of sync with actual practice as new AI impact assessments, triage results, and vendor reviews accumulate. This Skill catches that drift by comparing approved AI practices against the current policy, so the policy never silently misrepresents what the organization actually does. ## Core Features & Use Cases - Weekly Sweep Mode: Scans the outputs folder for AIAs, triage results, and vendor AI reviews since the last sweep, then classifies gaps as REQUIRED (policy contradicts practice) or ADVISABLE (policy is silent). - Direct Query Mode: Answers "we want to start doing X — does our AI policy cover it?" by diffing a proposed practice against policy commitments and the use case registry. - Drafted Policy Language: For every gap, quotes the current policy, explains the gap, and drafts suggested language matching the existing policy's style. - Use Case: A legal team approved an AI-assisted hiring screening tool via an AIA, but the AI policy still says "we do not use AI in employment decisions." The sweep flags this as a REQUIRED update and drafts replacement language before the misrepresentation becomes a liability. ## Quick Start Ask the AI to run a policy sweep of recent AI governance outputs, or describe a proposed new AI practice and ask whether the current AI policy covers it.
