verification-before-completion

Verification requires running fresh commands that prove your claim and capturing their full output and exit codes. Document what you ran, what the output shows, and confirm success explicitly—never skip this step based on confidence or perceived simplicity.

Valid evidence includes the exact command executed, complete output from that command, and exit code confirmation. Reproduce the original failure, apply your fix, then run the same verification command again to prove the issue is resolved.

Fresh verification captures real behavior in your current state. Automated checks may be stale or miss edge cases. Running commands yourself produces undeniable proof of success and prevents false positives that lead to broken trust and costly rework.

No. Before any completion claim, execute your test suite in the current environment and capture the output. Report the exit code and specific test results—not assumptions about what should pass.

Skipping verification due to confidence, running old commands instead of fresh ones, relying on partial output, or claiming success without checking exit codes. Enforce the rule: no completion without fresh, explicit evidence.

Include the exact verification command, full command output, exit code, and timestamp. This creates accountability and lets reviewers see proof independently, not just accept your assertion.