What problem does it solve?
It prevents false or premature “done” claims by forcing verification evidence before you commit, create a PR, or express satisfaction with work quality.
Core Features & Use Cases
- Verification-gated completion: Requires identifying the specific command that proves the claim, running the full command fresh, and confirming output and exit status before asserting success.
- Evidence-first status reporting: Guides you to state the actual status with evidence when verification fails, instead of using optimistic wording that implies passing.
- Common failure detection: Helps avoid typical traps like trusting prior runs, assuming partial checks are sufficient, or relying on an agent’s success message without independent validation.
- Use Case: Before merging code, you verify tests, lint/build, and any stated fixes with the exact commands whose output demonstrates the claim, then report results honestly.
Quick Start
Ask the AI to help you verify your change by stating the exact verification command you will run, then have it review the fresh command output and only allow you to claim completion if the evidence confirms it.