fixaiprompt
All templates
⚡ Advanced Techniques·Works on: claude, chatgpt, gemini

Generator–Verifier Pattern

Technique: Generator–Verifier

One model generates, another (or the same in a different role) verifies. Catches plausible-but-wrong output.

Advanced#verifier#agents#quality
0/2
Fill the template
2 placeholders left.
Live preview
Run this in Generator → Verifier mode.

Task:


Verification criteria (the verifier will check the generator's output against these):


**GENERATOR.**
Generate the output. Be confident; don't pre-hedge for the verifier. Return only the output.

**VERIFIER.**
Switch persona. You are now a separate, strict reviewer who didn't see the generation prompt. Apply each verification criterion to the generator's output and produce:
- PASS / FAIL on each criterion.
- For each FAIL: the specific evidence and the smallest possible fix.
- An overall verdict: SHIP / NEEDS REVISION / REJECT.

**REVISER (only if verifier said NEEDS REVISION).**
Apply ONLY the verifier's specific fixes. Do not change anything the verifier didn't flag. Output the revised version.

If SHIP: return generator output as-is.
If REJECT: return the verifier's evidence and recommend the user reformulate the task.
Inputs0 of 2
›See the lazy version this template replaces
Before — the lazy prompt
Do <task>.

Why it works

  • Splitting roles forces a true second-pass review instead of soft self-agreement.
  • Criterion-by-criterion PASS/FAIL with evidence is auditable — no vague 'looks good' verdicts.
  • The reviser's strict 'only fix what verifier flagged' rule prevents drift / over-rewriting.
  • Three-verdict system (SHIP/NEEDS REVISION/REJECT) catches the dangerous middle ground.

Make this one yours

Replace the bracketed placeholders, then paste into the Prompt Fixer to lint your customisation before hitting send.

Open Prompt FixerBrowse more templates