verification-before-completion

Verification before completion requires running the specific test, build, or QA command that confirms your work, reading its full output and exit code, and only then asserting success. This prevents false claims and ensures honesty about what actually passed.

Before committing, run the exact test that reproduces the bug and confirm it now passes, capturing the output as evidence. This enforces fresh verification rather than assumptions about the fix.

Common rationalizations like "should work now" or "I'm confident" bypass verification entirely. The skill counters this by mandating a strict 5-step process: identify the verification command, execute it fully, read output and exit code, verify results, and only then make any completion claim.

The skill applies to both local development and CI/CD workflows. Any software development task requiring fresh verification evidence—tests, linting, builds, QA checks—uses the same gate function before asserting success.

Skipping verification violates the skill's core enforcement: it prevents false claims of completion and ensures accuracy. Without running and confirming the verification command, you cannot honestly assert that work is done.