---
name: research-fact-check-claims-prompt
description: A prompt that lists every claim in a piece of writing, gives evidence for and against each with real sources, names logical fallacies, rates each claim from A to F, and sums up how far to trust it. Use when the user wants to fact-check claims, check if something is true, spot misinformation, or weigh both sides of an argument.
license: MIT
metadata:
  category: research-prompts
  tags: "research,fact-check"
  author: johnny-bravo
  version: "1.0"
  updated: "2026-10-31"
  status: published
  source_url: "https://github.com/danielmiessler/Fabric/blob/8ce0909d8622a61b6aaf539d467ba31bb7e0b6dd/data/patterns/analyze_claims/system.md"
  source_license: MIT
  copyright: "Copyright (c) 2012-2024 Scott Chacon and others"
  original_author: "Daniel Miessler and Fabric contributors (analyze_claims pattern)"
  modified: "Converted by TheSkillVault into a fill-in prompt: neutral wording, dropped political labels, added an 'unverified' option, steps, safety and done-when sections."
  imported_on: "2026-10-31"
---

# Research and fact-check claims (prompt)

## When to use
- Before you share, quote or act on a post, ad, report or argument.
- You want both sides of a claim with sources, not an opinion.
- Checking your own writing for claims you cannot back up.

## Steps
1. Paste the text into the prompt below, replacing {{TEXT}}.
2. Use an AI that can browse the web, so sources can be real and current.
3. Open at least one cited source for each claim you care about. Treat any link you cannot open as unverified.
4. Decide what you believe from the evidence, not from the rating alone.

## Prompt
```text
You are a neutral, careful analyst of claims and arguments.

1. ARGUMENT SUMMARY: the main argument in under 30 words.
2. For each factual claim in the text:
   - CLAIM: the claim in under 16 words.
   - SUPPORTING EVIDENCE: real, checkable facts or data with sources (title + link).
   - COUNTER EVIDENCE: real, checkable facts or data against it, with sources.
   - LOGICAL FALLACIES: any fallacies, each with a short quoted example.
   - RATING: A (definitely true), B (likely), C (mixed), D (unlikely), F (definitely false), or U (cannot verify).
3. OVERALL: the lowest, highest and average rating.
4. OVERALL ANALYSIS: 30 words on the argument's strengths, weaknesses, and how a reader should update their view.

Never invent a source. If you cannot find one, say "no source found" and rate the claim U.

TEXT:
{{TEXT}}
```

## How to adapt
- For health, legal or money claims, add "prefer official and peer-reviewed sources".
- For a quick check, ask only for the CLAIM and RATING lines.

## Gotchas
- AI models can make up sources. Every link must be opened before you trust it.
- This is not professional advice. Check medical, legal or financial claims with an expert.
- Works with any chat AI that can search the web. No tool lock-in.

## Done when
- Each claim has a rating and at least one source you have opened yourself, or is marked U.
- You can say in one line which claims hold up and which do not.

## Example prompt
"Is this viral post about screen time and sleep actually true? Check the claims."

---
Converted from the "analyze_claims" pattern in danielmiessler/Fabric (MIT, Copyright (c) 2012-2024 Scott Chacon and others). See LICENSE.txt.