verification-before-completion

To prevent false completion claims, enforce a strict evidence policy requiring fresh, verifiable command output before any success assertion is made. This ensures claims of passing tests, linting, or building are always backed by concrete verification evidence.

An AI claims a bug is fixed without running tests because it lacks a mandatory verification policy. Enforcing a rule that requires reviewing fresh test suite output before claiming completion prevents shipping unfixed bugs or incomplete features.

Evidence-based verification for code completion is a policy that mandates running verification commands and reviewing their output before asserting a task is complete. It guards against premature success claims by requiring fresh, verifiable proof across all development stages.

Mandatory verification works with linting and building processes by applying the same evidence policy across all development stages. It requires fresh command output from linting or building tools before any claims of success or completion are accepted.

To verify code quality before committing changes, run the test suite and check the output shows zero failures. Enforcing a strict verification process before making success assertions prevents the accidental introduction of regressions.

You should not use an automated verification policy when you cannot run fresh verification commands or review their output. The policy depends entirely on generating concrete evidence, making it unsuitable for environments where command execution is restricted.