skills.rest
BrunoAMSilva
BrunoAMSilvaCOMMUNITY
Explore all skills by this author →
GitHub

doublecheck

Extract verifiable claims from AI-generated text and verify them via web searches.

Updated Apr 14, 2026
Search Tags:#fact-checking#verification#web-search#hallucination-detection#verification-report#claims-extraction
One-click install
npx skills add https://github.com/BrunoAMSilva/my-config --skill doublecheck-brunoamsilva
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: doublecheck
Source: https://github.com/BrunoAMSilva/my-config/tree/main/coding/skills/doublecheck
Command: npx skills add https://github.com/BrunoAMSilva/my-config --skill doublecheck-brunoamsilva

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes assets (resource) components.

What problem does it solve?

Doublecheck reduces the risk of relying on hallucinated or unsupported AI assertions by extracting verifiable claims from model outputs, finding primary sources, and flagging suspicious or fabricated content so users can verify before acting.

Core Features & Use Cases

  • Claim extraction: Pulls every verifiable statement from target text and assigns tracking IDs for follow-up.
  • Source verification: Runs targeted web searches to find primary or authoritative sources and records URLs for each claim.
  • Adversarial review: Actively looks for hallucination patterns such as fabricated citations, precise unsourced statistics, temporal errors, and plausible-but-wrong associations.
  • Modes & outputs: Supports persistent active mode with inline verification summaries and one-shot mode that produces a full three-layer verification report using a provided template.
  • Use cases: Verify assistant responses before publishing, check legal or regulatory citations, audit research summaries, and validate factual claims in technical documentation.

Quick Start

Please verify the factual claims in the following assistant reply and produce an inline verification summary with sources when available.

Frequently Asked Questions about doublecheck

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

FAQPage Schema
How do I fact-check AI generated text for hallucinations and unsupported claims?▼

Fact-checking AI generated text for hallucinations involves extracting verifiable claims, running targeted web searches for primary sources, and flagging fabricated citations or temporal errors. This process produces inline verification summaries or full reports with source URLs.

What is the best way to verify factual claims in legal summaries and technical documentation?▼

Verifying factual claims in legal summaries requires extracting specific assertions and cross-referencing them against authoritative external sources. Adversarial hallucination checks identify plausible-but-wrong associations, ensuring technical documentation accuracy before publication.

How to extract verifiable claims from a document and cross-reference them with web sources?▼

Extracting verifiable claims from a document assigns tracking IDs to each statement, enabling targeted web searches for primary sources. This cross-referencing produces a three-layer verification report detailing source URLs and flagged hallucination patterns.

Can I check assistant responses for fabricated citations and precise unsourced statistics?▼

Checking assistant responses for fabricated citations uses adversarial review to detect precise unsourced statistics and temporal errors. It validates factual assertions against external sources, providing inline verification summaries to ensure reliability.

Does claim extraction and source verification work for research summaries needing factual validation?▼

Claim extraction and source verification works effectively for research summaries by pulling every verifiable statement and recording primary source URLs. It actively identifies hallucination patterns, providing a full verification report for factual validation.

When should I not rely on inline verification summaries for AI outputs?▼

Inline verification summaries should not be relied upon when a comprehensive audit trail is required. For legal or regulatory citations, use one-shot mode to generate a full three-layer verification report with source links instead of brief inline checks.

Related Skills

Explore More Content & Communication Skills→
ImaginerLabsImaginerLabs

doublecheck

Extract verifiable claims from AI-generated text and produce structured verification reports.

Official
Advanced
angryconsultantangryconsultant

doublecheck

Verify claims in AI outputs and surface credible sources.

Community
Advanced
shysssthanhtrishysssthanhtri

doublecheck

Extract verifiable claims from AI-generated text and verify them via web searches.

Community
Advanced
unfinishablemapunfinishablemap
3

pessimistic-review

Analyze site content for logical gaps and unsupported claims.

Community
Basic