verification-before-completion

Verify code changes by running the relevant testing, linting, or building commands and checking their actual output. You can only claim task completion after fresh verification evidence confirms zero failures, preventing premature success assertions.

An evidence before assertions policy mandates executing verification commands and confirming their output before reporting any work as done. This process ensures success claims for bug fixes or builds are always backed by tangible, freshly generated proof.

Yes, the verification process applies to all stages of development including bug fixing, linting, and building. It ensures that any claim of tests passing or bugs fixed is validated by executed command output before completion is reported.

Agents prematurely declare completion when lacking an enforced verification process. Without mandatory checks, success claims lack supporting evidence, which risks shipping broken work because tests were not actually run and confirmed.

Stop reporting incomplete bug fixes by enforcing a strict verification policy. You must run the relevant test command, examine its output, and only confirm the fix if the output shows zero failures before making any success claims.

The best way to ensure accurate assertions is to mandate running verification commands and confirming their output before claiming success. This strict process prevents shipping incomplete work by catching issues through testing before they are reported.