verification-before-completion

To prevent false completion claims in code review and commit gates, require fresh verification output before asserting success. Execute the full verification command in-session, paste the complete output, check for zero failures, and ban unverifiable phrases.

Evidence-first verification for pull requests mandates running tests and linters fresh, pasting the complete output and exit status, and explaining what the output proves before claiming completion.

Enforce a command-first workflow by mandating a specific message structure: run the verification command, paste complete output, explain what the output proves, then claim completion.

Agent delegations claim false success when they lack reproducible evidence. This is resolved by requiring fresh verification execution, pasting full output, checking exit codes, and integrating state-tracking and failure-recovery patterns.

Yes, you can use verification checks for linter and build verification in development workflows by running the verification command fresh, pasting the full output, and confirming a zero failure exit status before making any success claim.

Asserting bug fixes without fresh verification output leads to false completion claims. Limitations include unverifiable phrases, missing exit codes, and lack of failure-recovery patterns, which are banned to ensure honesty and reproducibility.