verification-before-completion

Verification before completion requires running your test suite fresh, reading the full output and exit codes, and confirming zero failures exist. Only then document the concrete evidence—command run, output captured, result confirmed—before claiming success or committing.

Use a five-step verification gate: identify the validation command, run it fresh, read complete output and exit codes, verify results match your claim, then state success with concrete evidence. This systematic approach catches assumptions before they become commits or merges.

Verification before completion applies to builds, linters, and PR status checks. Run each validation command fresh, capture full output, confirm exit codes indicate success, and document what you verified. This prevents premature PR creation based on assumptions.

Common failures stem from skipping fresh verification runs, ignoring exit codes, or rationalizing incomplete output. Verification before completion identifies these patterns by enforcing concrete evidence: actually run the command, read the complete result, and verify it against your claim before proceeding.

Yes. The five-step verification process works across any validation tool—test suites, build systems, linters, and PR status checks. Identify your specific verification command for each workflow stage, run it fresh, capture output, and verify results before asserting completion.