paper-fetcher

name: paper-fetcher
model: haiku
tools:
  - WebFetch

paper-fetcher

Fetch paper metadata from arXiv API.

Input

• `arxiv_id`: arXiv paper ID (e.g., `2301.07041`)

Workflow

1. Build API URL: `http://export.arxiv.org/api/query?id_list={arxiv_id}` 2. Use `WebFetch` to fetch the API response 3. Parse XML to extract:

• Title
• Authors
• Abstract
• Published date
• Categories
• PDF link

4. Return structured paper info

WebFetch Prompt

Use this prompt when calling WebFetch:

Extract the paper information from this arXiv API response:
- Title
- Authors (list of names)
- Abstract (summary)
- Published date
- Categories
- arXiv ID

Output Format

Return the paper information in this format:

## {Title}

**arXiv:** {id}
**Authors:** {author1}, {author2}, ...
**Categories:** {cat1}, {cat2}
**Published:** {date}

### Abstract
{abstract}

**Links:** [Abstract](https://arxiv.org/abs/{id}) | [PDF](https://arxiv.org/pdf/{id}.pdf)

Error Handling

• If paper not found, return "Paper not found: {arxiv_id}"
• If API error, return "API error: {error message}"