- Home
- Skills
- Research & Analysis
- Research & Source Credibility Gate Agent Skill
Works with the AI Agents you already use
Research & Source Credibility Gate Agent Skill
Adversarially audit research and AI claims for evidence gaps, source credibility, and framing bias.
$14
Research & Source Credibility Gate Agent Skill
Example session with this skill installed
Audit this claim: "Recent studies show that 85% of developers prefer remote work, citing a 2022 survey by TechCorp." Source link provided: https://techcorp.com/reports/dev-trends-2022.
- Read your context and instructions
- Compiled the research & source
I have completed the audit of the claim regarding developer work preferences.
Research Credibility Audit
1. Overall Verdict
- Verdict: VERIFY
- Risk level: Medium
- Review scope: Factual claim and the cited TechCorp 2022 report.
Reason: While the source exists, the specific percentage cited in the claim does not match the primary data in the report.
2. Findings
Finding 1 — Statistical Mismatch
- Verdict: REJECT
- Severity: High
- Exact passage: “Recent studies show that 85% of developers prefer remote work”
- Location: Paragraph 1, Sentence 1.
- Issue type: Citation mismatch.
- Why it matters: The 85% figure is the central evidence for the argument; an error here invalidates the premise.
Evidence checked: TechCorp Dev Trends Report 2022 (accessed via link). The report states that 65% prefer fully remote, while 20% prefer hybrid. The 85% figure is a combined total, not a "remote work" preference as strictly defined.
- What remains uncertain: Whether more recent 2023 or 2024 data supersedes this 2022 study.
- What to verify next: Locate 2024 developer survey data to see if the trend has shifted.
3. Source Quality Review
- Source: TechCorp 2022 Report
- Source type: Industry Survey (Secondary)
- Relevance: High
- Credibility: Reputable industry publisher, but sample size (n=500) is relatively small.
- Recency: Stale (2 years old).
4. Priority Verification Checklist
- Correct the 85% figure to specify it includes "Remote and Hybrid" preferences.
- Search for a 2024 primary source to confirm if the preference holds post-RTO mandates.
Next steps
- Update the text to distinguish between fully remote and hybrid percentages.
- Perform a search for "2024 developer sentiment reports" to update the recency of the claim.
Connects securely to your tools. The creator never sees your data.
About this skill
Research and AI-generated reports often suffer from hallucinations, circular citations, and cherry-picked data. This skill functions as an adversarial auditor, stress-testing claims against source quality to ensure your output is defensible. It moves beyond simple summarization to identify where evidence fails to support specific assertions.
What it does
- Claim extraction identifies discrete factual, numerical, and causal statements for independent analysis.
- Evidence auditing matches specific passages against cited sources to detect misrepresentations or weak support.
- Verdict assignment classifies findings as TRUST, VERIFY, or REJECT based on evidence strength and recency.
- Risk assessment weights findings by severity to determine if a report is safe for production or needs major revision.
- Framing analysis detects selective evidence, missing baselines, and correlation-causation fallacies.
How it works
- Intake the research brief, article, or AI response along with any provided source links or citations.
- Audit each material claim by inspecting source methodology, publication dates, and primary data.
- Generate a structured Credibility Audit report detailing specific contradictions or evidence gaps.
- Prioritize necessary verification actions to resolve high-severity risks.
Frameworks & tools
This skill is framework-agnostic and designed for use within LLM environments that support web browsing or document analysis. It operates on structured logic to evaluate research quality regardless of the underlying industry domain.
Why this beats prompting it yourself
Standard prompts often produce "yes-men" agents that confirm your existing text. This skill uses a strict adversarial workflow and a predefined verdict schema to force the agent to find friction points and evidence gaps it would otherwise ignore.
Use cases
- Auditing AI-generated technical whitepapers for hallucinated citations.
- Reviewing market research reports for stale data or one-sided framing.
- Fact-checking blog posts or articles before publication.
- Validating internal policy documents against source regulations.
Known limitations
Does not guarantee exhaustive fact-checking or replace expert judgment. Verdicts are limited by the accessibility of cited sources and the current knowledge cutoff of the agent.
Cannot verify claims if sources are behind paywalls or inaccessible. Does not replace legal or medical expert review. Limited by the accuracy of the provided text and source availability.
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 28 days ago
Permissions
No special permissions declared or detected