verification-before-completion

Identify the command that proves the claim, run it fresh and in full, read the complete output including exit codes and failure counts, then state the claim only with that evidence attached. Skipping any step counts as an unverified claim.

Only the actual test command output showing zero failures counts as evidence. Previous runs, partial checks, linter results, or statements like "should pass" are explicitly not sufficient under this verification gate.

No. Agent success reports must be independently verified by checking the version control diff and confirming the actual changes exist before reporting completion. Trusting the report alone is listed as a common failure.

No. A linter does not check compilation, so a clean lint result is not evidence of a successful build. You must run the build command and confirm exit code 0 before claiming the build passes.

Use the red-green cycle: write the test, run it to confirm it passes with the fix, revert the fix and confirm the test fails, then restore the fix and confirm it passes again. A test that only passes once is not verified.

Apply it before any success or completion claim, expression of satisfaction, commit, pull request, task handoff, or delegation to agents. The rule covers exact phrases, paraphrases, and any implication that work is correct or done.