verification-before-completion

To stop false completion claims, you need verification gatekeeping that forces the agent to execute specific test commands and analyze their output before reporting success. This enforces evidence-based assertions by requiring concrete exit codes or test results.

Evidence-based work completion is a verification protocol requiring objective analysis of command outputs like test results and diffs before declaring a task done. It prevents agents from assuming code works based on partial checks by demanding concrete proof.

To ensure pull request quality, enforce a verification protocol that runs the full test suite and confirms zero failures before submission. This requires the agent to execute testing commands and analyze the output for evidence-based reporting.

Yes, this verification approach applies to software development workflows including linting, building, and bug fixing. It requires executing specific verification commands for each task type and objectively analyzing their output to ensure reliability.

Agents report false success when they lack verification gatekeeping, leading to broken code. Enforcing evidence-based completion prevents this by requiring the execution of specific verification commands and objective analysis of their output before claiming success.