verification-before-completion

To prevent false completion claims during code review, require verification evidence by forcing the agent to run the exact proving command, read the complete output, and confirm the exit code before asserting any success.

Evidence-based verification in CI workflows is the practice of blocking success language until fresh execution of the exact proving command confirms test, lint, or build results with explicit output and exit status.

You enforce a verification gate by requiring the agent to run full test and lint commands, read complete output, confirm failure counts equal zero, and only then assert pull request readiness with explicit evidence.

Yes, this approach blocks success assertions when only partial checks, earlier runs, or speculative reports are used, prohibiting satisfaction language without fresh verification evidence from a complete command execution.

You should require fresh execution of the exact proving command because relying on earlier runs or "should/probably" thinking leads to false completion claims, whereas confirming exit codes and failure counts guarantees actual release readiness.

Yes, you can use this verification gate to confirm bug fixes by requiring the agent to run the full symptom regression command, check the complete output for expected behavior, and assert the fix with explicit evidence.