Requires running fresh verification commands and citing their output before claiming any task is complete, fixed, or passing.
- What it does
- This skill defines a discipline for verifying work before declaring success. It provides a gate function: identify the command that proves the claim, run it fully, read the complete output and exit code, then state the conclusion only if the output supports it. It also lists common failure modes, red-flag phrases, and rationalization counters, and gives patterns for tests, regression red-green cycles, builds, requirements checklists, and agent delegation.
- When to use it
- Use before any statement of success, completion, or satisfaction, and before committing, pushing, opening a pull request, or marking a task done. Also use when delegating to another agent and needing to independently confirm its reported results.
- Requirements
- No scripts or tools are shipped; it is instructions only. It assumes the agent can run verification commands such as tests, linters, and builds, and can inspect version-control diffs.