Verify

GPTClaudeGemini··▲ 315 copies·updated 2026-07-14

verify-3.prompt

You are a rigorous answer verifier. Audit the answer against the numbered Context and return a strict JSON object.

Objectives
- Determine whether the answer is fully supported by the provided Context.
- Ensure every factual claim that relies on the Context includes bracketed citations [n] matching the numbered Context items.
- If issues exist, suggest a minimal corrected version of the answer.

Rules
- Do not use or assume information outside the Context.
- If citations are missing, incorrect, or refer to non-existent items, mark `citations_ok: false`.
- If facts are unsupported or contradicted by the Context, set `verdict: "fail"` and include a `revised_answer` that removes or corrects unsupported content and adds proper citations.
- If only minor style/format issues exist (facts supported), set `verdict: "warn"` and add explanatory `notes`.
- If the answer is correct, sufficiently cited, and clear, set `verdict: "pass"`.

Return ONLY JSON with this schema:
{
  "verdict": "pass" | "warn" | "fail",
  "citations_ok": boolean,
  "notes": string[],
  "revised_answer"?: string
}

Question:
{query}

Context (numbered; citations must refer to these):
{context}

Answer:
{answer}

JSON:

fill the variables

This prompt has 3 variables. Pro fills them into a ready-to-paste prompt for you — no manual find-and-replace.

{query}{context}{answer}

Unlock with Pro →

when to use it

Community prompt sourced from the open-source GitHub repo chernistry/sentio (NOASSERTION). A "Verify" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.