evidence-audit

Verify numerical claims in an evidence HTML against PubMed abstracts via NCBI E-utilities.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/lucasmiachon-blip/OLMO --skill evidence-audit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: evidence-audit
Source: https://github.com/lucasmiachon-blip/OLMO/tree/main/.claude/skills/evidence-audit
Command: npx skills add https://github.com/lucasmiachon-blip/OLMO --skill evidence-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill automates the verification of numerical claims in a single evidence HTML by cross-referencing PubMed abstracts and ensuring consistency with cited data.

Core Features & Use Cases

  • Extract PMIDs and DOIs from an evidence HTML.
  • Cross-check numerical claims against PubMed abstracts using NCBI E-utilities.
  • Flag missing references and potential data fabrication for manual review.

Quick Start

Run the audit on a given HTML file path to obtain a structured report.

Frequently Asked Questions about evidence-audit

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

FAQPage Schema
How do I verify numerical claims in an evidence HTML against PubMed abstracts?

You can verify numerical claims by running this Skill to cross-reference evidence HTML data against PubMed abstracts using NCBI E-utilities. It performs identity checks for PMIDs and DOIs, returning a structured report with verification results and missing reference findings.

Can I use NCBI E-utilities to cross-check cited references without batch processing?

Yes, this Skill verifies references against PubMed abstracts using NCBI E-utilities without batch processing or external search. It extracts PMIDs and DOIs from your evidence HTML and flags missing references or potential data fabrication for manual review.

What is the best way to audit evidence HTML for data fabrication and missing references?

The best way to audit evidence HTML is to use this Skill to identify and verify all numerical claims against PubMed abstracts. It performs V1 identity checks for PMIDs, DOIs, and references, and V2 claim checks to flag potential data fabrication and missing references.

Does this PubMed abstract verification tool support batch processing of multiple evidence files?

No, this Skill does not support batch processing or external search. It is designed to verify numerical claims in a single evidence HTML against PubMed abstracts using only NCBI E-utilities, returning a structured report ready for pipeline orchestration.

How do I extract PMIDs and DOIs from an evidence HTML file for verification?

Extract PMIDs and DOIs by running the audit on a given HTML file path. The Skill automatically identifies these identifiers, performs V1 identity checks and V2 claim checks against PubMed abstracts, and returns missing reference findings in a structured report.

What limitations should I expect when verifying claims against PubMed abstracts using NCBI E-utilities?

Limitations include strict verification against abstracts only, no batch processing, and no external search capabilities. The Skill verifies a single evidence HTML at a time and flags missing references and potential data fabrication for manual review rather than auto-correcting them.