Verification Before Completion

Verification Before Completion

The Iron Law

NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE

If you haven't run the verification command in this message, you cannot claim it passes.

Violating the letter of this rule is violating the spirit of this rule.

The Gate Function

BEFORE claiming any status or expressing satisfaction:

1. IDENTIFY — What command/check proves this claim?
2. RUN — Execute it fresh and complete (use exec or read)
3. READ — Full output, check exit code, count failures
4. VERIFY — Does output actually confirm the claim?
   - If NO: State actual status with evidence
   - If YES: State claim WITH evidence
5. ONLY THEN: Make the claim

Skip any step = lying, not verifying.

Common Failures

| Claim | Requires | NOT Sufficient | |-------|----------|----------------| | "Tests pass" | Test command output: 0 failures | Previous run, "should pass" | | "Build succeeds" | Build command: exit 0 | Linter passing, "looks good" | | "Bug fixed" | Original symptom gone + test passes | "Code changed, assumed fixed" | | "File created" | read or ls confirms existence + content | "I wrote the file" | | "Config works" | Service restart + endpoint responds | "Config is valid JSON" | | "Subagent completed" | Verify actual changes (git diff, read files) | Trust subagent's "success" report | | "Requirements met" | Line-by-line checklist against spec | "Tests pass" |

Verification Checklist

For code changes:

• [ ] Code runs without errors (exec the test/build command)
• [ ] All tests pass (not just new ones)
• [ ] The specific requirement is demonstrably met
• [ ] No regressions in existing functionality

For file operations:

• [ ] File exists at expected path (read it back)
• [ ] Content matches intent
• [ ] Permissions correct if relevant

For deployments:

• [ ] Service is running
• [ ] Endpoint responds correctly
• [ ] Logs show no errors

For configuration:

• [ ] Config is valid (parse/lint it)
• [ ] Change survives restart
• [ ] Dependent services still work

Red Flags — STOP

• Using "should", "probably", "seems to"
• Expressing satisfaction before verification ("Great!", "Done!", "Perfect!")
• About to commit/push without verification
• Trusting subagent success reports without checking
• Relying on partial verification
• ANY wording implying success without having run verification

Rationalization Prevention

| Excuse | Reality | |--------|---------| | "Should work now" | RUN the verification. | | "I'm confident" | Confidence ≠ evidence. | | "Just this once" | No exceptions. | | "Linter passed" | Linter ≠ compiler ≠ runtime. | | "Subagent said success" | Verify independently. | | "Partial check is enough" | Partial proves nothing. |

The Rule

If you cannot verify a step, do not mark it complete. Note exactly what remains unverified and why. Partial completion honestly reported is better than false completion.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: gb3h
• Source: gb3h/agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.