跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
其他技能

arxiv-submission

Prepare and post preprints to arXiv without processing failures or leaks, covering TeX source packaging (.bbl, figures, case-sensitive paths), stripping private comments, choosing categories and license, endorsement, metadata (title, abstract, comments field), versioning and replacements, and linking the journal DOI later. Use when posting a paper to arXiv, when arXiv's TeX compilation fails, before uploading source, or when updating a preprint after acceptance. Includes a zero-dependency preflight checker.

Kalaris Labs (Sayan Chowdhury) · 已有 AI 审核记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add KalarisLabs/research-agent-skills --skill arxiv-submission

技能说明


name: arxiv-submission description: Prepare and post preprints to arXiv without processing failures or leaks, covering TeX source packaging (.bbl, figures, case-sensitive paths), stripping private comments, choosing categories and license, endorsement, metadata (title, abstract, comments field), versioning and replacements, and linking the journal DOI later. Use when posting a paper to arXiv, when arXiv's TeX compilation fails, before uploading source, or when updating a preprint after acceptance. Includes a zero-dependency preflight checker. license: MIT compatibility: Python 3.9+ standard library for the preflight script; a local TeX installation to build the .bbl. metadata: version: "1.0" category: journal-formats maintainer: Kalaris Labs author: Kalaris Labs (Sayan Chowdhury) tags: arXiv, preprint, LaTeX, submission, open access

Posting to arXiv

Verify first. Policies (size limits, accepted formats, license options, endorsement, moderation rules) change. Read https://info.arxiv.org/help/submit/index.html and https://info.arxiv.org/help/submit_tex.html before each submission. Also check the target journal's or conference's preprint policy, especially anonymity periods for double-blind venues.

1. Build clean source

  1. Compile locally from scratch (latexmk -pdf main.tex) and confirm the PDF is final.
  2. Copy only what is needed into a fresh folder, or run Google's arxiv_latex_cleaner (pip install arxiv-latex-cleaner; arxiv_latex_cleaner paper/), which strips comments and unused files.
  3. Run the preflight:
python scripts/arxiv_preflight.py paper_arxiv/

It flags: missing .bbl (upload the compiled .bbl named like the main file; with biblatex the .bbl must match arXiv's biber version, so plain BibTeX .bbl files are safer), case-mismatched or absolute paths, EPS with pdflatex, build artifacts, hidden files, private files (response letters, reviews, drafts), TODO/reviewer comments (source is public forever), unicode or spaces in file names, and large uploads.

  1. Put \pdfoutput=1 in the first lines of the main file for pdflatex processing.
  2. Compress: photos as JPEG, huge vector plots rasterized at 300 dpi, no unused figures.

2. Metadata

FieldGuidance
TitleSame as the paper. Plain text. Use TeX only where needed for math
AuthorsFull names in order. Match the paper and your other papers (affects author pages)
AbstractPlain text, no citations or \cite. Keep math minimal. Lines are rewrapped
Comments"12 pages, 5 figures; accepted at XYZ 2026" and code link if public
Primary categoryThe single best fit (e.g. cs.LG, q-bio.GN, astro-ph.CO); cross-lists sparingly
Journal-ref / DOIAdd after publication (you can update metadata without a new version)
LicenseChoose deliberately: it cannot be made more restrictive later. CC BY is common for OA mandates. Check funder and journal requirements

First submission in a category may require endorsement from an established author.

3. After submitting

  • Check the compiled PDF in the preview. arXiv's TeX Live may differ from yours.
  • Announcement follows the daily schedule. Fix problems before the cutoff with "replace" or "unsubmit".
  • Replacements create v2, v3...: summarize changes in the comments. Old versions stay public.
  • After journal acceptance, update to the accepted manuscript if the publisher's policy allows (check the self-archiving policy, e.g. via Sherpa Romeo), and add the DOI.
  • Withdrawal does not delete earlier versions. Never post content you may not share (patient data, embargoed results, third-party figures without permission).

Related skills

ml-paper-writing (templates), venue-templates, bibtex-hygiene, citation-verification, reproducibility-statement (code/data links).