arxiv-reader

Read and analyze arXiv papers by fetching their public LaTeX source. Converts LaTeX into clean text suitable for LLM analysis.

Description

This skill fetches arXiv papers from the public arXiv API (arxiv.org), flattens LaTeX includes, and returns clean text. No local file access is required — all content is fetched over HTTPS from arXiv's public endpoints and cached in memory for the session.

Network access: Only connects to arxiv.org and export.arxiv.org to download publicly available paper source tarballs and metadata. No other network connections are made. No data is sent to external services — this is read-only.

Caching: Results are cached in memory (process-scoped) for fast repeat access within the same session. No files are written to disk.

Usage Examples

• "Read the paper 2301.00001 from arXiv"
• "What sections does paper 2405.12345 have?"
• "Get the abstract of 2312.09876"
• "Fetch paper 2301.00001 without the appendix"

Process

1. Quick look — Use arxiv_abstract to get a paper's abstract before committing to a full read
2. Survey structure — Use arxiv_sections to understand the paper's outline
3. Deep read — Use arxiv_fetch to get the full flattened LaTeX for analysis

Tools

arxiv_fetch

Fetch the full flattened LaTeX source of an arXiv paper.

Parameters:

• arxiv_id (string, required): arXiv paper ID (e.g. 2301.00001 or 2301.00001v2)
• remove_comments (boolean, optional): Strip LaTeX comments (default: true)
• remove_appendix (boolean, optional): Remove appendix sections (default: false)
• figure_paths (boolean, optional): Replace figures with file paths only (default: false)

Returns: { content: string, arxiv_id: string, cached: boolean }

Example:

{ "arxiv_id": "2301.00001", "remove_appendix": true }

arxiv_sections

List all sections and subsections of an arXiv paper.

Parameters:

• arxiv_id (string, required): arXiv paper ID

Returns: { arxiv_id: string, sections: string[] }

Example:

{ "arxiv_id": "2301.00001" }

arxiv_abstract

Extract just the abstract from an arXiv paper.

Parameters:

• arxiv_id (string, required): arXiv paper ID

Returns: { arxiv_id: string, abstract: string }

Example:

{ "arxiv_id": "2301.00001" }

Notes

• Results are cached in memory — repeat requests within the same session are instant
• Paper IDs support version suffixes (e.g. 2301.00001v2)
• Very large papers may take 10-30 seconds on first fetch
• arxiv_abstract uses the public arXiv Atom API for fast metadata retrieval
• No filesystem writes — all caching is in-memory only
• Only connects to arxiv.org (read-only, public data)