- Home
- Skills
- Research & Analysis
- AI Output Evidence Checker
AI Output Evidence Checker
Analyze AI-generated text to extract factual claims and generate a structured verification ledger to prevent hallucinations.
$5
Works with the AI Agents you already use
AI Output Evidence Checker
Example session with this skill installed
Fact-check this snippet: "A 2022 study by the Global Tech Institute found that 68% of developers prefer Rust for system-level tasks due to its memory safety features, which reduced crashes by 40%."
- Read your context and instructions
- Compiled the ai output evidence
I have analyzed the provided text for factual claims and potential hallucinations.
[AI Evidence Check Report]
📊 Summary:
- Total Claims Found: 3
- High Risk / Needs Check: 2
📋 Claim Ledger:
1.
Claim: "A 2022 study by the Global Tech Institute found that 68% of developers prefer Rust for system-level tasks."
- Status: [HIGH RISK]
Verification Action: Search Google for "Global Tech Institute 2022 Rust study". Note: The name 'Global Tech Institute' is a common hallucination pattern for generic-sounding organizations.
2. Claim: "Rust's memory safety features reduced crashes by 40%."
- Status: [NEEDS CHECK]
Verification Action: Search for: "Rust memory safety impact on crash rates 40% statistic source". Verify if this figure comes from a specific company report (e.g., Microsoft or Google) rather than a general study.
3. Claim: "Rust features memory safety for system-level tasks."
- Status: [VERIFIED]
Verification Action: None required. This is a core architectural feature of the Rust programming language documented in the official Rust Book.
[Self-Check]
- All factual claims extracted
- Confidence/Verification flags assigned
- Actionable verification steps provided for risky claims
- No blind trust or generic "looks good" statements
Next steps
- Search for the "Global Tech Institute" to confirm if the organization actually exists.
- Cross-reference the 40% crash reduction stat with the 2022 Stack Overflow Developer Survey or Microsoft's security blog.
- Replace the high-risk citations with verified data from the 2023 "State of Rust" report.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Large Language Models frequently hallucinate specific statistics, case studies, and citations that look plausible but are entirely fabricated. Developers and content teams risk reputational damage if they publish AI-generated technical reports or articles without a rigorous, claim-by-claim verification process.
What it does
- Deconstructs AI-generated text into a structured "claim ledger" of verifiable facts.
- Identifies high-risk patterns common in hallucinations, such as fake academic papers or fabricated quotes.
- Assigns specific verification statuses (VERIFIED, NEEDS CHECK, HIGH RISK) to every extracted claim.
- Generates precise, actionable search queries for human auditors to use for final validation.
Why this beats prompting it yourself
Generic prompts often result in the AI simply agreeing with its own previous output or providing vague "looks good" feedback. This skill enforces a "No Blind Trust" rule, requiring a formal ledger output and preventing the model from skipping the granular analysis required for true fact-checking.
Use cases
- Auditing AI-generated research summaries before sharing with clients or stakeholders.
- Vetting blog posts and marketing copy for accurate statistics and industry references.
- Verifying technical documentation or whitepapers created via LLM workflows.
Known limitations
This tool identifies claims and suggests verification steps but cannot browse the live web to confirm truth in real-time. It requires a human to execute the suggested search queries.
Requires human execution of suggested search queries. It flags potential hallucinations but does not have a real-time live-web browsing connection for final verification.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 months ago
Permissions
No special permissions declared or detected