Hallucination Risk Reviewer

Classify AI output claims and generate hallucination risk reports.

9|1|Updated Mar 15, 2026

Search Tags:#calibration #risk-assessment #hallucination #LLM evaluation #risk-review #claim-types

Ready to run

npx skills add https://github.com/Notysoty/openagentskills --skill hallucination-risk-reviewer

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: Hallucination Risk Reviewer
Source: https://github.com/Notysoty/openagentskills/tree/main/skills/hallucination-risk-reviewer
Command: npx skills add https://github.com/Notysoty/openagentskills --skill hallucination-risk-reviewer

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill reviews an LLM-generated response or the output of an AI-powered feature to assess hallucination risk, identify claim types, and provide mitigations, delivering a structured risk assessment to guide usage decisions.

Core Features & Use Cases

Identify claim types: Factual world claims, citations, procedural/instructional content, numerical data, and domain-expertise assertions.
Assess risk by claim type and calibration issues, producing a risk score and recommended mitigations.
Generate a formal risk assessment suitable for human review or integration into product workflows.
Use cases include evaluating AI support responses, content generation, and high-stakes decision prompts.

Quick Start

Analyze the provided AI output for hallucination risk and generate a structured risk assessment.

Frequently Asked Questions about Hallucination Risk Reviewer

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

FAQPage Schema

How do I assess hallucination risk in LLM outputs for high-stakes decisions?▼

Assess hallucination risk in LLM outputs by classifying claims into factual, citation, procedural, numerical, and domain-expertise categories. This generates a structured risk report with high/medium/low scores, calibration issues, and mitigation guidance for human review.

What is the best way to identify claim types that cause AI hallucinations?▼

Identify hallucination claim types by categorizing AI outputs into factual world claims, citations, procedural content, numerical data, and domain-expertise assertions. This classification quantifies propagation risk and produces actionable mitigations for AI support bots and content generation.

Can I evaluate AI support bot responses for citation and numerical hallucinations?▼

You can evaluate AI support bot responses for citation and numerical hallucinations by reviewing the generated output. The skill classifies claim types, assesses calibration issues, and delivers a formal risk assessment with an overall usage recommendation.

Does this hallucination risk review work with AI outputs from Claude Code and Cursor?▼

This hallucination risk review works with AI outputs from Claude Code, Cursor, Codex, and similar environments. It analyzes the generated text to quantify propagation risk and provides structured mitigation guidance suitable for integration into product workflows.

How do I generate a formal risk assessment for AI-generated content?▼

Generate a formal risk assessment for AI-generated content by analyzing the output for hallucination risks and calibration issues. The process produces a structured report with high, medium, and low risk categories alongside an overall recommendation for usage decisions.

When should I perform a hallucination risk assessment on AI procedural instructions?▼

Perform a hallucination risk assessment on AI procedural instructions when evaluating outputs for high-stakes decisions or content generation. This identifies procedural claim inaccuracies, evaluates propagation risk, and provides mitigations to ensure safe usage.