confirm-criterion

Confirm Criterion Skill

This tool converts a self-claimed pass into a formal pass, unblocking can_complete_goal. It MUST be called by an independent reviewer agent — not the agent that performed the work.

Who Should Call This

• ✅ A subagent spawned specifically to review the work
• ✅ A different agent session that re-verifies independently
• ❌ The same agent that called validate_criterion with passed

What to Do

1. Read the criterion description and its original evidence
2. Independently re-verify — do NOT trust the original evidence:
   • If method=command: re-run the command yourself
   • If method=file: read the file yourself and check the content
   • If method=url: make the HTTP request yourself
3. If the re-verification passes, call confirm_criterion with YOUR fresh evidence (not a copy of the original)
4. If the re-verification fails, call validate_criterion with status=failed and your evidence showing the failure

Evidence Type

Must be high-confidence: command, file, or url. text evidence is rejected — a reviewer saying "looks fine" without re-running is worthless.

Example

{
  "criterion_id": "api-200",
  "evidence": "reviewer re-ran: curl -s localhost:3000/health → 200 {\"status\":\"ok\"}",
  "evidence_type": "command"
}

Rejection Scenarios

Scenario	Result
Criterion is not passed	Error: not a self-claimed pass
Criterion is already confirmed	Error: not a self-claimed pass
evidence_type=text	Error: high-confidence evidence required
Empty evidence	Error: evidence is required