Exa MCP Server
Lets your AI agent search the web and read pages as clean text, without an account
by Exa
arXiv MCP Server
by Joe BlazickConnectorResearch something
Lets your AI agent search arXiv, read papers section by section and export BibTeX
About arXiv MCP Server
A connector that runs on your computer and lets an AI agent search arXiv, download papers, read them section by section, export BibTeX citations and keep watches on topics for new papers.
For StudentsResearchersPeople following a research field
Made by Joe Blazick
An open-source project by developer Joe Blazick, published on GitHub.
- Stars
- ★ 3.2k
- License
- Apache-2.0
- Cost
- Free
- Level
- Intermediate
- Runs
- Local
Overview
arXiv MCP Server is a connector that runs on your computer and gives an AI agent tools for arXiv papers. Your agent can search by topic, author or category, download a paper, read its outline and one section at a time, export BibTeX citations and follow references through Semantic Scholar. You can also save topic watches and check them later for new papers. It suits students and researchers who keep up with papers in their field.
It reads papers one section at a time, including the authors' original LaTeX, so the agent can quote a specific part instead of skimming a whole PDF.
Try these first
Start with the first one: it is a small task you can check easily.
- Get the abstract and BibTeX for arXiv paper [arXiv ID].First try · read-only
- Search arXiv for papers on [topic] in cs.CL from the last six months and list the five most relevant.
- Download [arXiv ID], show its outline, then read only the methods section.saves to your disk
- Watch the topic [topic] and tell me next week which new papers appeared.saves a watch
Example prompts written by Agent Flows Hub.
Capabilities
From the creator's documentation.
| Toolset | What your agent can do | Access |
|---|---|---|
| Search | Search arXiv by query, category, date and sort order | Read |
| Paper reading | Download a paper as Markdown, then read its outline and one section at a time | Read |
| LaTeX source | Read the authors' original LaTeX, one section at a time | Read |
| Citations | Export BibTeX and fetch references and citing papers from Semantic Scholar | Read |
| Topic watches | Save topics on your computer and check them for new papers | Read + write |
| Semantic search | Search downloaded papers by meaning; needs the optional pro install | Read |
Good to know
You want an agent to find recent papers on a topic and quote from the right section. The server searches arXiv, stores papers on your disk and returns bounded sections and BibTeX built from arXiv metadata.
Limits
- Search, source downloads and the citation graph call arXiv and Semantic Scholar; reading saved papers and topic watches stay on your computer.
- Install it from PyPI with uvx: the README warns that an unrelated npm package uses the same name.
Notes by Agent Flows Hub (original project name: arxiv-mcp-server), written from the creator's documentation, not from running it ourselves. arxiv-mcp-server README
Stuck setting it up?
Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.
More to research something
Other connectors
- Asana MCP Lets your AI assistant look up, create and update tasks and projects in Asana
- Atlassian MCP Server Lets your AI agent find, summarize and update Jira work items and Confluence pages
- Attio MCP Lets your AI assistant look up and update contacts, deals and notes in your Attio CRM
- Brave Search MCP Server Lets your AI agent search the web, news, images and videos through Brave Search
- Browser Harness Lets your coding agent use your own Chrome, signed in as you, to click, type and download
- Canva AI Connector Lets your AI agent create, edit, search and export designs in your Canva account
See all 52 connectors · An MCP server is missing or disconnected