home / mcp / arxiv mcp server

ArXiv MCP Server

Provides access to arXiv search, download, and read capabilities via an MCP interface with local storage.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "huanongfish-arxiv-mcp": {
      "command": "uv",
      "args": [
        "tool",
        "run",
        "arxiv-mcp-server",
        "--storage-path",
        "/path/to/paper/storage"
      ],
      "env": {
        "ARXIV_STORAGE_PATH": "~/.arxiv-mcp-server/papers"
      }
    }
  }
}

The ArXiv MCP Server lets AI assistants search arXiv and retrieve papers through a simple MCP interface. It stores downloaded papers locally for fast access and provides a straightforward set of tools to search, download, list, and read papers, all via MCP clients.

How to use

You interact with the ArXiv MCP Server through an MCP client. Use the four core tools to search for papers, download them, list what you have stored, and read the full text content of a downloaded paper. Each tool is designed to be called by your MCP client to perform a specific action without exposing you to protocol details.

How to install

Prerequisites: you need a runtime environment and the tooling described below to set up the ArXiv MCP Server.

Install via Smithery to automatically wire the client for Claude Desktop:

npx -y @smithery/cli install arxiv-mcp-server --client claude

Manual installation with uv for local running and development setup:

Clone the project and set up a Python virtual environment for development and testing:

# Clone and set up development environment
git clone https://github.com/blazickjp/arxiv-mcp-server.git
cd arxiv-mcp-server

# Create and activate virtual environment
uv venv
source .venv/bin/activate

# Install with test dependencies
uv pip install -e ".[test]"

Configuring MCP connections

Configure the MCP client to connect to the ArXiv MCP Server using a stdio (local) process. The server runs as a local tool and is started by invoking uv with the arxiv-mcp-server entry point and a storage path for downloaded papers.

{
  "mcpServers": {
    "arxiv_mcp_server": {
      "command": "uv",
      "args": [
        "tool",
        "run",
        "arxiv-mcp-server",
        "--storage-path", "/path/to/paper/storage"
      ]
    }
  }
}

Storage and environment

Papers you download are stored locally to speed up future access. The default storage location is configurable through an environment variable.

ARXIV_STORAGE_PATH=~/.arxiv-mcp-server/papers
``()

Available tools

search_papers

Search for arXiv papers with optional filters such as query text, max results, date range, and categories.

download_paper

Download a paper by its arXiv ID to local storage for offline access.

list_papers

List all papers currently stored locally on the server.

read_paper

Read the content of a locally stored paper by its arXiv ID.