arxiv 2 tools

MCP server exposing 2 tools for arxiv.

Endpoint

This URL is a JSON-RPC 2.0 endpoint over HTTP. Issue POST requests with a JSON-RPC body. Browsers and search crawlers land here on GET.

POST https://gateway.pipeworx.io/arxiv/mcp
Content-Type: application/json

{"jsonrpc":"2.0","id":1,"method":"tools/list"}

Tools

• search_papers — Search arXiv preprints. Plain text searches all fields; use prefixes for targeted queries: au:hinton (author), ti:transformer (title), abs:diffusion (abstract), cat:cs.AI (category), all:quantum (any field). Combine with AND/OR/ANDNOT, e.g., "ti:transformer AND cat:cs.LG". Returns id, title, authors, abstract, categories, published date, PDF URL.
• get_paper — Fetch a single arXiv paper by its ID (e.g., "2310.06825", "2310.06825v2", or legacy "cs.CL/0301001"). Returns full metadata: title, authors, abstract, categories, DOI (if linked), PDF URL.

Full documentation

Code samples (curl / TypeScript / one-click client install), schemas, and the live playground are on the pack page:
https://pipeworx.io/packs/arxiv/

About Pipeworx

Pipeworx is an open MCP gateway connecting AI agents to live data. pipeworx.io