doubao-reference-audit

Audit academic paper citations for authenticity, bibliographic accuracy, and claim-evidence support.

2|Updated Aug 9, 2026
One-click install
npx skills add https://github.com/DeepJH/doubao-skill-and-info --skill doubao-reference-audit-deepjh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doubao-reference-audit
Source: https://github.com/DeepJH/doubao-skill-and-info/tree/main/skills/doubao-reference-audit
Command: npx skills add https://github.com/DeepJH/doubao-skill-and-info --skill doubao-reference-audit-deepjh

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve? Academic papers often contain fabricated references, incorrect bibliographic details, mismatched in-text citations, and citations that do not actually support the claims they accompany. This Skill systematically audits a paper's entire reference apparatus and produces a professional, actionable citation review report. ## Core Features & Use Cases - Full Citation Mapping: Builds a complete map of in-text citations and bibliography entries, detecting missing, unused, duplicated, or ambiguous references. - Bibliographic Verification: Verifies titles, authors, years, venues, DOIs, versions (preprint vs. published), corrections, and retractions using Crossref and authoritative sources via the bundled lookup script. - Claim-Evidence Comparison: Judges whether each cited source genuinely supports the sentence it is attached to, flagging misquotation, overgeneralization, causal overreach, and secondary citation. - Report Generation: Produces a structured Markdown audit report from the included template and optionally creates a Feishu document. - Use Case: Before submitting a journal manuscript, provide the PDF or LaTeX source and receive a formal citation review identifying fabricated references, wrong DOIs, and claims that exceed what the cited studies actually show. ## Quick Start Audit the references in my attached paper and generate a citation review report checking authenticity, bibliographic accuracy, and whether citations support the claims.

Frequently Asked Questions about doubao-reference-audit

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

FAQPage Schema
How do I check if references in my paper are real?

Provide the paper or reference list and the audit verifies each entry against Crossref, publisher pages, arXiv, PubMed, and other authoritative sources. The bundled lookup_metadata.py script batch-queries Crossref by DOI or title to retrieve bibliographic candidates for verification.

How to check whether citations actually support my claims?

The audit reads each citation's context, classifies its role, and compares the paper's statement against what the source actually demonstrates. It flags partial support, overgeneralization, causal overreach, and secondary citation with specific revision suggestions.

What input formats does reference auditing accept?

The audit accepts PDF, Word, Markdown, LaTeX, plain reference lists, or readable paper text. For batch bibliographic checks, prepare a UTF-8 JSON list of references with optional DOI, title, author, and year fields.

Does the Crossref lookup script decide if a reference is correct?

No. The script only retrieves metadata candidates from the Crossref API and caches results. Matching and judging semantic support remain the auditor's responsibility; an ok status or high similarity score does not prove a reference is genuine or correctly cited.

When should I not use citation auditing?

It is not suited for open-ended literature reviews, single-paper summarization, ghostwriting, or pure language polishing. It requires an actual paper or reference list to examine and does not replace data, statistical, or ethics review.