AI Use Case Intake
Turns an AI/ML feature proposal into an attorney-ready intake record with a risk tier.
Getestet · Funktioniert
Was es kann
Legal-ops skill that walks through a proposed or changed AI/ML use case (model, data, outputs, affected persons, markets, vendors) and produces a structured intake record, a Low/Medium/High triage signal, and routing recommendations to specialist reviews (vendor terms, model risk, privacy). Triggers when a team needs legal/compliance/governance sign-off on a new or modified AI feature; explicitly defers to sibling skills for vendor-contract review or deep model-risk analysis.
Testbericht
Ran the same fully-specified AI-hiring-tool scenario through a no-skill baseline and the skill: the baseline confidently asserted 'NYC Local Law 144 ... almost certainly applies and the company is not currently compliant' with no verification, exactly the kind of unearned legal conclusion the skill is built to prevent; the skill output instead produced a structured, routable intake table with [CONFIRM]/[verify jurisdiction] placeholders and a clear attorney-escalation flag, no false-confidence claims.
Getestet am: 2026-07-16 · Claude Code 2.x (agent harness)
Installation
git clone https://github.com/zgbrenner/agentcounsel mkdir -p ~/.claude/skills cp -r agentcounsel/skills/ai-governance/ai-use-case-intake ~/.claude/skills/ai-use-case-intake
Befehle & Beispiel-Prompts
/ai-use-case-intakeTurns an AI/ML feature proposal into an attorney-ready intake record with a risk tier.
Skills reagieren auf normale Anfragen — keine Slash-Befehle nötig. Nach der Installation aktivieren Prompts wie diese den Skill (auf Englisch):
We're deploying an AI resume screener, get this ready for legal reviewTriage this new AI feature's risk level before we ship it company-wideCreate an intake record for this ML use case so legal can sign off on it