Prepare an arXiv submission with AI
Inspect a preprint’s source package, check its metadata and find material that should stay out of a public submission.
Prepare an arXiv submission with AI when a PDF builds on your machine but the source folder still contains working files. The arxiv-submission skill guides an agent through TeX packaging, metadata and version choices. Its local preflight script flags potential problems for the author to investigate before uploading anything.
A project might contain a figure whose filename differs from the reference by one capital letter, or an old response to reviewers beside the final manuscript. A clean build from a separate submission directory can reveal the first problem; checking the package contents helps prevent the second becoming public.
What it does
- Check figure references against the files in the proposed upload.
- Find stray drafts, private comments and review material that need attention.
- Verify the bibliography artefact needed to rebuild the manuscript.
- Compare the title, author order and abstract with the submission metadata.
- Prepare a replacement version and review the publication history implications.
Install and use
- Use Python 3.9+ and your authorised source folder for the checker. A local TeX installation supports a clean build; an arXiv account is needed to submit.
- Ask the agent to inspect a clean copy and explain any warnings without editing the original.
- Build again, review the current submission rules and confirm that all content may be posted.
FAQ
Source
Made by Kalaris Labs (Sayan Chowdhury). Original skill: arxiv-submission, published on GitHub under the MIT licence.
View on GitHub