What files are required for this skill to work?

It primarily requires an output/CLAIMS.md file containing the assertions to be audited against the manuscript.

How does it handle claims where evidence cannot be located?

It marks the claim as 'evidence not locatable' and generates a request for a clearer source pointer or re-extraction of the original claim.

Can it be used for non-academic projects?

Yes, it is highly effective for auditing technical documentation, white papers, and any project where claims must be backed by specific data or theory.

Does this skill write the missing evidence for me?

No, it is designed with a guardrail to only identify gaps and suggest fixes; it does not introduce new arguments or write the manuscript content.

What is the main purpose of the Evidence Auditor skill?

It systematically checks every claim in a document against the provided evidence to identify gaps, weak baselines, or unsupported conclusions.

Research Evidence Auditor

Name: Research Evidence Auditor
Author: WILLOSCAR

byWILLOSCAR

•

データサイエンスとML

Audits research claims against supporting evidence to identify gaps, unsupported assertions, and weak evaluation protocols.

The Research Evidence Auditor is a specialized skill designed to enhance the rigor of technical and academic writing by systematically cross-referencing claims with available evidence. It iterates through a list of extracted claims, evaluates empirical data—such as datasets, baselines, and evaluation protocols—and flags missing or weak justifications in a structured output file. By providing actionable fixes for every identified gap, this skill helps researchers and developers move away from hollow writing and ensure that every assertion in their manuscript or documentation is grounded in verifiable artifacts.

主な機能

01Structured validation of datasets, baselines, and evaluation protocols

02Actionable 'minimal fix' recommendations for missing evidence

0389 GitHub stars

04Severity classification for identified evidentiary weaknesses

05Systematic gap analysis for empirical and conceptual claims

06Guardrails to prevent the introduction of new claims or hollow discourse

ユースケース

01Auditing white papers for unsupported marketing or technical assertions

02Conducting internal peer reviews of research papers prior to submission

03Validating technical documentation claims against actual implementation benchmarks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add willoscar/research-units-pipeline-skills evidence-auditor

For use in Claude.ai and ChatGPT

Download Skill