verification-before-completion

To prevent false claims of code completion before a commit, enforce running exact verification commands, inspecting full output, and checking exit codes to provide explicit evidence before declaring success.

Evidence-first verification in CI and pre-commit workflows is the principle of executing tests, reading exit codes, and observing outputs before making any positive statement about work status or correctness.

To ensure regression testing passes before creating a pull request, identify the proving command, run it fully, verify outcomes, and report the actual state with observed evidence before marking tasks done.

You need to verify agent-reported results before accepting them to avoid regressions and broken releases, requiring fresh observable verification evidence from full command runs before accepting any reported success.

When verification commands fail before a merge, report the actual state with evidence instead of claiming success, ensuring no work is marked complete without passing exit codes and output inspections.

Yes, you can use this verification gate for local test runs, linters, commits, and CI builds, requiring full execution and output inspection before any success statements are made.