verification-before-completion

To enforce verification before a commit, run the full test suite and linter, inspect the entire command output and exit code, and include explicit evidence of the results before claiming completion.

The best way to prevent false completion claims is applying a verification gate that requires running full checks, reading exit codes, and reporting concrete evidence like pass counts rather than relying on memory.

Checking exit codes and full output is necessary because build commands can fail silently or produce partial successes, requiring explicit evidence from the terminal output to accurately assert success.

Yes, automated verification applies to agent-produced changes by requiring the agent to run regression checks and linters, inspect the full command output, and provide explicit evidence before reporting success.

Enforcing a verification gate adds a necessary validation step to the workflow, requiring developers and reviewers to execute commands and review output, which prevents premature claims but ensures code reliability before completion.