arXiv MCP Server

This MCP server wraps the arXiv API into a Claude-friendly interface, letting you search academic papers and pull abstracts without leaving your chat window. It handles the standard search operators (author, title, category) and returns structured metadata you can actually work with. I've used it to build quick literature reviews when scoping out a research angle, and it's faster than tab-switching to arXiv's web interface every time you need to check if someone's already published your idea. The real workflow win is chaining searches with summarisation. Ask Claude to find papers on a topic, then have it pull the abstracts and synthesise common themes. You can filter by date, sort by relevance or submission time, and limit result counts to avoid drowning in output. The PDF download feature is there if you need full text, though you'll still need to handle the actual reading yourself. Quirks: it's a thin wrapper, so you inherit arXiv's search quirks. Boolean operators work, but the syntax isn't always intuitive if you're used to Google Scholar. The server doesn't cache results, so repeated searches hit the API each time. That's fine for ad-hoc queries but less ideal if you're iterating on the same search terms. Also, no citation export built in, so you'll need to format references manually or pipe the metadata elsewhere. Skip this if you're doing deep bibliometric analysis or need advanced filtering beyond what arXiv's native search offers. It's not a replacement for Zotero or Mendeley. But if you're a researcher who lives in Claude and wants to shortcut the "open browser, search, copy-paste abstract" loop, it's a clean addition to your MCP stack. Installation is trivial, no API keys required, and it works out of the box with Claude Desktop.