What problem does it solve?
This Skill solves the problem of time-consuming, inconsistent manual screening of scientific papers for potential research misconduct, including image reuse, manipulated experimental data, and suspicious numeric patterns, by standardizing evidence collection and analysis for reviewers, editors, and investigators.
Core Features & Use Cases
- Automated PDF Processing: Converts scientific paper PDFs or public URLs to structured Markdown, tables, images, and metadata using MinerU, preserving all original content and precise location references.
- Citeable Evidence Ledger: Builds a unified, location-tagged evidence index of all text, tables, figures, captions, and table cells with page numbers, bounding boxes, and original values for full traceability.
- Deterministic Numeric Forensics: Runs automated checks for repeated values, suspicious terminal digit patterns, fixed table column relationships, and Benford's Law applicability to flag data anomalies without AI-generated guesswork.
- Annotated Evidence Images: Generates deterministic, source-linked PNG annotations for high-risk source data tables to support audit reports and manual review.
Use case: A journal editor can use this Skill to screen submitted manuscripts for data anomalies and image duplication before sending them for peer review, reducing the risk of publishing flawed research.
Quick Start
Use the paper-fraud-auditor skill to audit the scientific paper PDF at '/path/to/paper.pdf' for potential research integrity anomalies including image reuse, suspicious data patterns, and table inconsistencies.
