verification-before-completion

To enforce verification before completion, you must run the full verification command, review its complete output, and document exit codes and failures to ensure success claims are evidence-based across commits and pull requests.

Evidence-based completion requires running full verification commands and documenting outputs, exit codes, and failures before stating success, preventing unverified claims and eroding trust in collaborative development environments.

Claiming completion without verification risks inaccuracies because unverified success claims lack documented evidence like output logs and exit codes, which erodes trust among collaborators relying on genuine quality assurance.

You document verification outcomes by running the designated verification command, capturing the complete results, and recording the specific outputs, exit codes, and any failures before stating that the agent outputs or commits are successful.

Yes, evidence-based verification applies across commits, pull requests, and agent outputs by requiring the full verification command to run and its complete results to be reviewed, serving as guardrails for existing workflows.