verification-before-completion

To prevent false completion claims before opening a pull request, require fresh verification evidence by identifying the exact proof command, running it fully, and reading the complete output and exit status before asserting any done or passing status.

Evidence-based verification for release readiness is the practice of blocking premature satisfied phrasing like done or fixed until CI checks, lint, build, and test verification commands are fully executed and their results are confirmed in the same message.

To ensure your CI checks actually passed before declaring a bug fixed, you must run the full verification command, read the complete output, confirm the exit status, and only then make assertions that match the freshly verified results.

No, you cannot use partial test results to claim a code change is passing. Relying on partial checks, extrapolation, should pass language, or unverified agent reports are detected and blocked as typical shortcuts to prevent trust loss.

You need to run full verification commands for TDD workflows whenever you are about to communicate that tests pass, a build succeeded, or a bug is fixed, ensuring success statements are backed by real, freshly collected results.

Relying on unverified agent reports causes completion-safe communication failures because it allows teams and agents to claim work is finished without freshly collected verification evidence, directly increasing false-completion and trust loss.