Agent Output Review
Use this skill to catch the failure modes that appear when an AI agent produces something plausible but incomplete. The goal is not to be harsh; it is to protect the user from shipping ghost features, fake integrations, missing instructions, and brittle assumptions.
What To Review
Review the output against the user's actual goal, not just surface quality:
- Does it solve the requested problem?
- Does it work in the target environment?
- Are there fake, stubbed, or de
[Description truncada. Veja o README completo no GitHub.]