An Agent Skill that reviews agent work on behalf of the user.
Acts as the user's proxy when reviewing plans, work output, or completion claims from other agents. Evaluates against:
- Completeness checklist — post-deployment, global installs, AGENTS.md regeneration, delivery chain
- Thoroughness checks — item-level analysis, evidence-backed claims, rule compliance
- Known error patterns — shallow analysis, premature claims, missing post-deployment, stale state
Returns APPROVE or FLAG with specific concerns.
- Reviewing an agent's plan before approval
- Reviewing completed work before accepting
- Evaluating completion claims
npx skills add metyatech/skill-user-proxy --yes --globalMIT