Skip to main content
Claiming work is complete without verification is dishonesty, not efficiency. The verification-before-completion skill enforces a single rule: run the command, read the output, then claim the result.

​
When this skill fires

The skill description reads: “Use when about to claim work is complete, fixed, or passing, before committing or creating PRs — requires running verification commands and confirming output before making any success claims; evidence before assertions always.” It fires before:
  • Any variation of success or completion claims
  • Any expression of satisfaction (“Done!”, “Perfect!”, “Great!”)
  • Committing, creating PRs, or marking tasks complete
  • Moving to the next task
  • Delegating to agents

​
What it does

The skill enforces a gate function that must be completed before any claim: 
BEFORE claiming any status or expressing satisfaction:

1. IDENTIFY: What command proves this claim?
2. RUN: Execute the FULL command (fresh, complete)
3. READ: Full output, check exit code, count failures
4. VERIFY: Does output 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

​
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.

​
Common failures

ClaimRequiresNot sufficient
Tests passTest command output: 0 failuresPrevious run, “should pass”
Linter cleanLinter output: 0 errorsPartial check, extrapolation
Build succeedsBuild command: exit 0Linter passing, logs look good
Bug fixedTest for original symptom: passesCode changed, assumed fixed
Requirements metLine-by-line checklist against planTests passing
Agent completedVCS diff shows changesAgent reports “success”

​
Key patterns

✅ [Run test command] → [See: 34/34 pass] → "All tests pass"
❌ "Should pass now" / "Looks correct"

​
Red flags — stop

Stop before making any claim if you notice:
  • Using “should”, “probably”, or “seems to”
  • Expressing satisfaction before verification
  • About to commit or create a PR without running verification
  • Trusting an agent’s success report without checking the diff
  • Relying on a partial verification
  • Thinking “just this once”

​
Rationalization prevention

ExcuseReality
”Should work now”Run the verification
”I’m confident”Confidence is not evidence
”Just this once”No exceptions
”Linter passed”Linter is not a compiler
”Agent said success”Verify independently
”I’m tired”Exhaustion is not an excuse
”Partial check is enough”Partial proves nothing

​
Example scenario

You’ve just fixed a bug in the authentication service. Before saying “fixed”, the verification-before-completion skill requires you to:
  1. Identify the verification command: npm test src/auth/
  2. Run it: npm test src/auth/
  3. Read the output: 18/18 tests passed, 0 failures
  4. Verify: yes, the output confirms the claim
  5. Make the claim: “Bug fixed. npm test src/auth/ shows 18/18 tests passing including the new regression test for empty email validation.”
Not: “Should be fixed now” or “I think that covers it.”

Test-driven development

Provides the tests that verification-before-completion runs to confirm work is done.

Systematic debugging

After a fix, verification-before-completion confirms the fix actually resolved the issue.

Build docs developers (and LLMs) love

Get started for freeTalk to us