validate-answer

name

validate-answer

description

Trust-check an answer, plan, or decision you already have — multiple identical independent reads converge per question, agreement signals confidence and splits flag what's contested, then walk each decision. TRIGGER when: user says 'validate', 'trust-check', 'is this right', 'pressure-test my decision'; user has a near-final answer or a few focused decisions to confirm. SKIP when: no answer yet, want one built (best-answer); surfacing what's missing (find-gaps).

Validate Answer

Trust-check an answer or a few focused decisions.

When to use

You have an answer, plan, or decision and want to know if you can trust it.

• Multiple identical independent reads: agreement = confidence, disagreement = contested. The reviewers are deliberately the same — diversity here would destroy the signal.
• NOT this if you have no answer yet and want one built → best-answer.
• NOT this if you want to surface what you missed → find-gaps.
• NOT this if you're grading one concrete proposal → second-opinion.

Steps

Step 1 — Dispatch

• Reviewers: R0 (you, concurrent) + R1, R2 (two general-purpose subagents in parallel).
• Prompt (identical for all three): artifact path(s), the questions, a confidence (0.00–1.00) and impact score per recommendation.
• Preflight: if any question is vague, sharpen via AskUserQuestion first.
• Output schema per question: recommendation, confidence (0.00–1.00), impact — render Label (value): Minimal (0.25) · Low (0.5) · Medium (1) · High (2) · Massive (3) — 1-2 sentence reason.
• Unprompted observations (any reviewer): 1-sentence claim, raising reviewer(s), confidence 0.00–1.00, impact; merge duplicates.