verify-before-claim

# Verify Before Claim **NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE.** ## Usage Template **Prompt** ```text Use verify-before-claim. Before saying this is done, identify the proof command, run it fresh, read the output, and state the actual status with evidence. ``` **Use Case** - Preventing unsupported claims that code, docs, tests, or workflows are fixed, passing, complete, or working. **Expected Result** - The agent reports command output, exit code, pass/fail count, and any remaining unverified risk. **Output Example** - `Command: npm test`, `Exit code: 0`, `Result: 42 passed, 0 failed`, plus residual risks. **Verification Case** - No completion language appears unless fresh evidence is shown in the same response. **Verified Effect** - The agent shifts from confidence-based status claims to evidence-based status reports. ## Success Metrics - A fresh verification command or inspection is run in the current session. - Output includes command, exit code or observable evidence, result, and residual risk. - No completion claim is made when evidence is missing, stale, or failing. ## When to Use - Before saying "done", "fixed", "complete", "passing", "working" - Before committing or creating a PR - Before expressing satisfaction with results - When the user asks "is it working?" - Any time the agent feels the urge to say "it should work" ## The Gate Function ``` BEFORE claiming any status: 1. IDENTIFY: What command proves this claim? 2. RUN: Execute t...