fact-checker

Extract claims from Markdown, PDF, and DOCX documents and validate them against web sources.

1|Updated Feb 16, 2026
One-click install
npx skills add https://github.com/dobachi/claude-skills-marketplace --skill fact-checker-dobachi
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fact-checker
Source: https://github.com/dobachi/claude-skills-marketplace/tree/main/plugins/fact-checker/skills/fact-checker
Command: npx skills add https://github.com/dobachi/claude-skills-marketplace --skill fact-checker-dobachi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires puppeteer, node, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Fact-checker helps you catch inaccuracies in AI-generated writing by extracting checkable claims and validating them against real, accessible sources.

Core Features & Use Cases

  • Claim extraction from documents: Ingests Markdown/plain text or reads files (including PDF/DOCX) and converts them into normalized plain text for analysis.
  • Automated, parallel verification: Verifies each claim using web search plus headless-browser retrieval (Puppeteer), and runs multiple checks concurrently for speed.
  • Structured Markdown reporting: Produces a comprehensive fact-check report with verdicts per claim, including evidence and explanations.
  • Structured Checklist Mode: When you need thorough auditing (e.g., link validity, numeric accuracy, reliability tiers), it applies dimension-based checks defined in references/workflow.md.
  • Batch URL fetching: Uses batch fetching to validate many cited links at once and capture HTTP status and final redirect targets.

Quick Start

Ask the assistant: “Fact-check this document and verify all cited references; then output a detailed Markdown report.”

Frequently Asked Questions about fact-checker

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify claims and cited references in an AI-generated document?

To verify claims in AI-generated documents, this Skill extracts checkable statements and validates them against web sources using parallel search and Puppeteer retrieval, outputting a structured Markdown fact-check report with evidence and verdicts.

Can I fact-check content from PDF and DOCX files?

Yes, fact-checking supports PDF and DOCX inputs by reading files and converting them into normalized plain text before extracting claims and validating them against accessible web sources.

How do I perform batch URL validation for many cited links at once?

Batch URL validation is performed using concurrent fetching to check many cited links simultaneously, capturing HTTP status codes and final redirect targets for comprehensive reference auditing.

Does claim verification require Puppeteer and Node to run?

Yes, automated claim verification requires Node and Puppeteer to run headless-browser page retrieval, enabling parallel web searches and structured checklist audits for link validity and numeric accuracy.

What's the best way to audit numeric accuracy and link validity in a Markdown report?

The best way to audit numeric accuracy and link validity is using the Structured Checklist Mode, which applies dimension-based checks to evaluate reliability tiers and produce a detailed Markdown fact-check report.

Why does web citation verification sometimes fail to validate certain claims?

Web citation verification may fail if claims lack accessible online sources or if headless-browser retrieval encounters blocked content, preventing the extraction of sufficient evidence for a definitive verdict.