verification-before-completion

You can enforce verification by running specific test, lint, and build commands, then objectively analyzing their output to confirm the work state before claiming completion.

AI agents often claim work is finished due to rationalization, using excuses like "it should work" instead of running validation checks. Enforcing a mandatory verification gate with explicit command execution prevents this failure.

The best way to validate build status before a pull request is to run the full test suite and build command, ensuring the code is truly ready for integration by checking fresh exit codes and system states.

To prevent rationalization in development workflows, implement a verification gate that requires the execution of specific commands and objective analysis of their output to confirm work state, rejecting unfounded confidence.

Evidence-based work completion applies to all development workflows including linting, testing, building, and agent delegation by enforcing rigorous verification protocols and checking exit codes before finalizing any report.