| name | council-review |
| description | Run an Agent Review Council over a product, workflow, codebase, architecture, prompt, agent package, or GTM artifact. Produces evidence-backed structured findings, scorecard, risks, decision record, and validation plan.
|
| allowed-tools | Read, Grep, Glob, Bash |
Council Review Skill
Use this skill for full ARC runs that need coordinated evidence collection, independent reviewer passes, adjudication, and schema-valid outputs.
Workflow
- Confirm target, objective, council definition, constraints, and required outputs.
- Read
references/run-workflow.md for the end-to-end council procedure.
- Read
references/output-contract.md before creating or validating run artifacts.
- Read
references/external-reviewers.md only when the run includes Codex, GitHub, Copilot, LangGraph, or other non-Claude reviewers.
- Write artifacts under
runs/<date>-<slug>/.
- Validate with
uv run arc validate runs/<date>-<slug>.
Ground Rules
- Findings without evidence are hypotheses.
- High-severity findings require strong evidence or explicit uncertainty.
- Run independent reviewer passes before synthesis.
- Preserve accepted, rejected, disputed, and watchlist findings.
- Do not create tickets or durable memory without approval.
- Keep final output concise, but preserve the structured artifacts.
