arXiv Paper and Source Downloader

Problem

Reading arXiv papers usually means handling several small tasks at once: extracting an ID from an abs/pdf URL, downloading the PDF or LaTeX source, and saving multiple papers with clean filenames. Manual copying and renaming becomes error-prone when titles contain :, /, or ?, or when titles are very long.

How It Works

The skill is built around scripts/download_arxiv.py, using only the Python standard library. It supports:

• Single download: provide an arXiv ID, arxiv.org/abs/ID, or arxiv.org/pdf/ID; saves the PDF and renames it after the paper title
• Keyword search: queries arXiv API results, then lets the user choose IDs to download
• Source download: uses --source to fetch the .tar.gz LaTeX package
• Batch download: processes multiple IDs in one command and reports success/failure counts

The workflow extracts IDs or URLs, determines the output directory, runs the script, and returns the saved path. Filename collisions get (1) or (2) appended; long titles are truncated at word boundaries; batch requests are spaced by 0.5s.

Notes

Not every paper has an available source, and arXiv may return 403. Metadata requests time out at 30s, downloads at 120s. If export.arxiv.org is blocked, retrying or changing the network may be necessary.