Stop shipping hallucinations. Review AI output before it goes live.
AI generates confident-sounding text that is often wrong. It doesn't hedge. It doesn't flag uncertainty. It presents fabricated citations, flawed logic, and invented statistics with the same polished tone it uses for accurate information.
This matters now more than ever:
- 30% of new code at Amazon and Google is AI-generated (internal reports, 2025)
- AI-written marketing copy, business proposals, and emails are going out daily with zero quality review
- LLMs hallucinate at rates between 3% and 27% depending on the task and model
- A single uncaught error in a client deliverable, a contract, or production code can cost thousands
The gap isn't generation. It's review. Most teams have no systematic process for checking AI output before it ships.
This repo contains 5 ready-to-use review prompts from the full 35-prompt AI Output Quality Checker. Each prompt is a structured review template with context variables you fill in — paste your AI-generated content, run the prompt, and get a detailed quality report.
| # | Prompt | Category |
|---|---|---|
| 1 | Factual Accuracy Audit | Research & Analysis |
| 2 | Tone and Voice Check | Content & Copy |
| 3 | Logic and Reasoning Review | Research & Analysis |
| 4 | Code Security Scan | Code Review |
| 5 | Email Professionalism Check | Email & Communication |
The complete AI Output Quality Checker covers every type of AI output you need to review:
| Category | Description | Prompts |
|---|---|---|
| Content & Copy | Marketing copy, blog posts, social media, ad text | 7 prompts |
| Code Review | Security, performance, readability, dependency risks | 6 prompts |
| Business Docs | Proposals, SOWs, contracts, reports, slide decks | 6 prompts |
| Research & Analysis | Fact-checking, source verification, reasoning, data claims | 6 prompts |
| Email & Communication | Professional tone, clarity, intent alignment, follow-ups | 5 prompts |
| Meta-Review | Review-of-reviews, consistency checks, pre-publish gate | 5 prompts |
Choose the review prompt that matches what you need to check. Each file tells you when to use it.
Every prompt has [BRACKET VARIABLES] — replace them with your specific content and context.
Paste the completed prompt into any capable LLM (Claude, GPT-4, Gemini, etc.) along with the AI-generated content you want reviewed.
Using the Factual Accuracy Audit prompt:
[AI OUTPUT] = A blog post your AI wrote about GDPR compliance
[DOMAIN] = EU data privacy law
[INTENDED AUDIENCE] = Small business owners in the US
[CRITICALITY LEVEL] = High — this will be published on our company blog
The review prompt will return a structured report flagging unsupported claims, outdated references, misleading generalizations, and confidence ratings for each factual assertion.
The complete AI Output Quality Checker includes:
- All 35 review prompts across 6 categories
- Every category fully covered (not just samples)
- Pre-publish quality gate system — a final meta-review workflow that runs before anything goes live
- Scoring rubric for each review type (pass / flag / fail thresholds)
- Chaining guide — how to run multiple review prompts in sequence for high-stakes output
- Priority prompt updates as models evolve
Run a quality check right now without downloading anything.
Try the free demo on Hugging Face Spaces →
By COCÓ — Smart Tools & Products | Good Stories LLC
COCÓ builds practical AI tools for professionals who use AI daily and need their output to be reliable. We focus on the gap between generating AI content and shipping it with confidence.
This free starter kit is released under the MIT License.
Copyright 2026 Good Stories LLC. See LICENSE for details.
The free prompts in this repository may be used, modified, and shared freely. The full 35-prompt pack is a separate commercial product available on Gumroad.