| name | arxiv-reader |
| description | Reads and extracts content from Arxiv HTML papers for Q&A. Use when the user shares an Arxiv HTML link (arxiv.org/html/...) or asks about a paper from Arxiv. |
| allowed-tools | WebFetch |
Arxiv Paper Reader
When to Activate
Activate when the user:
- Shares a link matching
arxiv.org/html/* - Asks to read, summarize, or discuss an Arxiv paper
- References a paper ID like
2512.09742
Instructions
- Use WebFetch to retrieve the paper from the Arxiv HTML URL
- Extract: title, authors, abstract, and all sections
- Provide a 2-3 sentence summary of the paper
- Indicate readiness for follow-up questions
Answering Questions
- Quote relevant passages when appropriate
- Reference specific sections or figures
- Distinguish between what the paper states vs. interpretation
- If the paper doesn't address a question, say so clearly
URL Format
Arxiv HTML links follow the pattern: https://arxiv.org/html/{paper_id}
If user provides abstract URL (arxiv.org/abs/...), convert to HTML format by replacing /abs/ with /html/.