verification-before-completion

Code completion verification requires running specific commands like linters and tests, then examining their full output and exit codes. This ensures concrete evidence precedes any assertion of work being complete.

Evidence-based verification mandates presenting concrete command output and exit codes as proof of success. It prevents premature claims of bug fixing or passing status by ensuring actual results confirm the assertion before communication.

Agents claim task completion without test output when a strict verification process is missing. Enforcing mandatory verification requires examining actual testing and building command results before any success assertion is communicated.

Yes, automated verification prevents premature claims of passing status by mandating specific commands for testing, linting, and building. It requires zero failures in the output before any success declaration is communicated.

Relying on assumptions for code review verification leads to costly errors and dishonest claims. The limitation is that previous states do not confirm current success; you must run validation commands and present actual output showing zero failures.