verification-before-completion

To prevent false claims of task completion before a pull request, enforce an evidence-before-assertions policy requiring fresh command execution and output verification for all success claims. This ensures build and test validations are independently executed rather than assumed.

AI reports bug fixes as complete without running tests due to premature assertions based on assumptions. Enforcing mandatory verification requires independent execution of test suites and checking exit code 0 to confirm the specific fix addresses the reported issue with evidence.

Mandatory verification for code quality is a strict policy requiring executed verification commands and their output before claiming any task, test run, or build is successful. It prevents false positives by stopping status reporting based on past runs or incomplete checks.

To verify a successful build before committing code, execute the build command independently and confirm it exits with code 0. This strict verification process ensures all claims of passing tests or requirement fulfillment are backed by fresh execution evidence.

Yes, you can use automated verification to stop false positives in test runs by enforcing a policy that requires independent command execution and output validation. This prevents the AI from reporting success based on assumptions or incomplete checks before stating a task is complete.

Evidence-based verification for completion claims should not be skipped when committing code, fixing bugs, or creating pull requests. There are no safe exceptions to bypassing execution checks, as assuming success without fresh output risks integrating false positives and unresolved issues.