ArXiv

CLI and MCP server for core arxiv services

python

1

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "liamconnell-arxiv_for_agents": {
      "command": "uv",
      "args": [
        "run",
        "arxiv-mcp"
      ]
    }
  }
}

You can use the arXiv MCP server to empower your LLM assistant or automation with direct access to arXiv search, metadata, and PDF downloads. The MCP server exposes four tools for discovery and management and can run either as a local stdio process or via a remote HTTP endpoint, enabling seamless integration with your agent workflows.

How to use

Set up your MCP client to connect to the arXiv MCP server. You can run the server locally and have your agent send requests to it, or point a remote client at a reachable MCP endpoint. The four built-in tools cover searching papers, retrieving metadata, downloading PDFs, and listing downloaded files. Use these tools to build conversational flows that discover relevant arXiv content, fetch detailed information, and manage local PDFs for offline review.

How to install

Prerequisites you need before starting the MCP server:

Install the MCP server runtime using the command runner shown here.

Run the MCP server using one of the available entry points.

Running the MCP server locally (stdio)

# Option 1: Using the script entry point (recommended)
uv run arxiv-mcp

# Option 2: Using the module
uv run python -m arxiv.mcp

What you can do with the MCP server

search_papers — search arXiv with field-specific queries and boolean operators
get_paper — retrieve complete metadata for a specific paper and optionally download its PDF
download_paper — download the PDF for a given paper and return the local file path
list_downloaded_papers — list all locally downloaded PDFs with sizes and paths

Available tools

search_papers

Search arXiv with advanced query syntax, including field prefixes (ti:, au:, abs:, cat:), boolean operators, pagination, and sorting. Returns paper metadata such as title, authors, abstract, categories.

get_paper

Get detailed information about a specific paper by ID, arXiv:ID or similar formats. Optionally downloads the PDF and returns complete metadata including DOI, journal references, and comments.

download_paper

Download the PDF for a specific paper to the local .arxiv directory. Returns the file path and size, and supports force re-download.

list_downloaded_papers

List all locally downloaded PDFs with IDs, sizes, and file paths to help you manage your collection.