doublecheck

To verify AI-generated text, this Skill extracts verifiable claims, locates supporting or contradicting sources via web search, and runs adversarial review to flag hallucination patterns. It delivers a structured verification report for human review.

Three-layer verification applies self-audit, source verification, and adversarial review to provided text. It identifies verifiable claims, assigns stable IDs for tracking, and produces a structured report to guide risk assessment and decision-making.

Provide the text you want verified and the Skill runs multi-layer verification to generate a detailed report. It operates in inline verification mode by default, extracting claims and returning mapped IDs within the text.

Yes, the Skill extracts verifiable claims from AI outputs and assigns stable IDs for tracking. This allows you to map specific statements to their sources and review them individually in the structured verification report.

Yes, the adversarial review layer detects hallucination patterns, inconsistencies, and risky claims. It assesses retrieved supporting sources through automated web searches to flag potential issues in the AI-generated text.

Automated source verification retrieves and assesses supporting sources via web search but requires human review of the final structured report. It flags hallucination patterns and risky claims to guide risk assessment, but does not replace human decision-making.