skills.rest
mrsonord2240
mrsonord2240COMMUNITY
Explore all skills by this author →
GitHub

arxiv-preflight

Scans LaTeX manuscripts for LLM artifacts, placeholder content, and unverifiable references before arXiv submission.

Updated Sep 15, 2026
Search Tags:#latex#bibtex#preflight#arxiv#research integrity#reference verification#llm artifacts
One-click install
npx skills add https://github.com/mrsonord2240/openscience-specialists --skill arxiv-preflight-mrsonord2240
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: arxiv-preflight
Source: https://github.com/mrsonord2240/openscience-specialists/tree/main/specialists/manuscript-revision-specialist/versions/1.0.0/package/skills/arxiv-preflight
Command: npx skills add https://github.com/mrsonord2240/openscience-specialists --skill arxiv-preflight-mrsonord2240

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve? Manuscripts prepared with AI assistance can contain leftover LLM meta-comments, hallucinated references, placeholder data, and missing AI-use disclosures that trigger arXiv rejection or author bans. This Skill runs a structured submission-readiness preflight that locates concrete, citable evidence of these problems and produces a fix list with a PASS / PASS_WITH_FIXES / HOLD decision. ## Core Features & Use Cases - AI-artifact scanning: Detects LLM meta-comments, prompt residue, chat tokens, placeholders (TODO, XX%), and AI tools listed as authors using a data-driven regex pattern library. - Reference verification: Parses BibTeX, checks structural validity, and cross-references entries against Crossref, arXiv, OpenAlex, and Semantic Scholar to flag fabricated or mismatched citations. - Structured reporting: Merges extraction, scan, and verification outputs into a single Markdown report with severity-graded findings, file:line locations, and a human sign-off checklist. - Use Case: Before uploading a multi-file LaTeX paper to arXiv, run the preflight to catch an "As an AI language model" sentence left in methods.tex, an unverifiable citation in refs.bib, and an unfilled XX% placeholder in the results table. ## Quick Start Run an arXiv preflight check on my LaTeX project directory and tell me whether the manuscript is submission-ready with a list of any blocking issues.

Frequently Asked Questions about arxiv-preflight

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

FAQPage Schema
How do I check a LaTeX paper before submitting to arXiv?▼

Run the extraction script on your LaTeX project directory, then the artifact scanner and reference verifier, and finally generate the merged Markdown report. The report gives a PASS, PASS_WITH_FIXES, or HOLD decision with file:line locations for every finding.

How to detect hallucinated or fake references in a BibTeX file?▼

The verifier parses each BibTeX entry, checks DOI and arXiv ID structure, then queries Crossref, arXiv, OpenAlex, and Semantic Scholar in order. Entries with no match across all reachable services plus formatting tells are graded BLOCKER as likely fabricated.

Does the preflight work on PDF files instead of LaTeX source?▼

Yes, but PDF input relies on pdftotext and only recovers raw text and numbers; section, citation, and label structure is lost. LaTeX source is strongly preferred because reference and cross-reference checks are far more reliable.

What happens when reference verification has no network access?▼

The verifier marks the reference section INCOMPLETE and records which services were unavailable. The overall decision cannot be PASS; it becomes PASS_WITH_FIXES at best with an explicit instruction to rerun verification with network access.

Can this tool detect whether a paper was written by AI?▼

No, it deliberately does not judge AI authorship from writing style. It only flags concrete evidence such as LLM meta-comments, leaked prompt text, chat tokens, placeholder data, and AI tools listed as authors.

What are the limitations of the arXiv preflight check?▼

It does not perform full semantic peer review, automated claim verification, plagiarism detection at scale, or predict arXiv acceptance. Claim and number consistency checks surface candidates for the author to confirm rather than verifying them automatically.

Related Skills

Explore More Education & Research Skills→
tqLi99tqLi99

paper-write

Convert PAPER_PLAN.md and NARRATIVE_REPORT.md into a compilable LaTeX manuscript.

Community
Advanced
Pythagora-ioPythagora-io
9

qa-phase4

Validate Phase 3 completion, coordinate sign-off, and generate an HTML report with evidence.

Official
Advanced
Imad-OuteImad-Oute

scientific-visualization

Create publication-ready scientific figures with multi-panel layouts and journal-compliant exports.

Community
Intermediate
sparkfabriksparkfabrik
3

glab

Automate GitLab issues, merge requests, CI pipelines, and repository operations via the glab CLI.

Official
Intermediate