verification-before-completion

Prevent false completion claims by enforcing a 5-step verification gate that requires running full test commands, reading complete output, and confirming actual results before asserting any work is done. This eliminates shipped bugs caused by unverified success statements.

Verification evidence required before creating a pull request includes running the full test suite to confirm zero failures and validating the build succeeds. You must read complete command output and match it against your claims before issuing any positive status statements.

Agents report bug fixes as complete without confirmation due to rationalization, skipping verification by assuming changes are correct. A mandatory verification gate prevents this by requiring fresh evidence from running actual checks and validating the original symptom is gone before any fix claim is made.

Yes, you can use automated verification for regression testing and agent delegation tasks by applying a 5-step gate function. It requires identifying the proof command, running it fully, reading complete output, verifying against the claim, and stating the result for all development workflows.

The limitations of relying on code quality assumptions include false completion claims, shipped bugs, missed requirements, and broken trust. Without running full verification commands and reading complete output, rationalization bypasses actual testing and leads to unvalidated success assertions.