verification-before-completion

To enforce a verification protocol before claiming code completion, mandate running specific verification commands and examining their output and exit codes to present evidence before asserting any task is finished or passing.

Premature work completion declarations cause quality issues by allowing incomplete or broken code to ship. Enforcing evidence-based verification prevents this dishonesty and ensures all necessary checks are performed before success claims.

The best way to verify code readiness before a commit or PR creation is to run linters and test suites, verify no errors are reported, and present the command output as evidence before stating the code is ready for review.

Yes, you can use automated verification to prevent shipping broken code by mandating that specific verification commands run successfully and their exit codes are checked before any success claims are made during development stages.

To provide evidence-based proof that a code fix is passing, run the required verification commands, examine the output and exit codes, and present this tangible proof before asserting that the task is finished or fixed.

You need to run verification commands during all stages of development and task completion, including code commits, PR creation, and agent delegation, to ensure work quality and prevent premature success claims.