Tool to work with arXiv, provide LLM with ability to search and read papers from there
Configuration
View docs{
"mcpServers": {
"andybrandt-mcp-simple-arxiv": {
"command": "python",
"args": [
"-m",
"mcp_simple_arxiv"
]
}
}
}You can run an MCP server that provides access to arXiv papers through their API to enable intelligent clients to search, filter, and retrieve paper metadata, abstracts, and full text in Markdown. This server is designed for easy local or remote use and can be integrated with MCP-enabled clients to streamline scholarly paper discovery and retrieval.
You will run the MCP server locally or on a network and connect your MCP client to it. The server supports searching arXiv by title and abstract, filtering by submission date, sorting by submission date, update date, or relevance, and viewing the total count of matching results. For each paper, you can access metadata, abstracts, and links to available formats, along with the full paper text converted to Markdown.
Prerequisites: Python must be installed on your system. You may also use a containerized approach, but this guide focuses on running the server directly with Python.
Manual installation steps:
pip install mcp-simple-arxiv
# Run the MCP server as a local stdio service using Python
# You will connect via an MCP client by specifying the Python module as the entry point
python -m mcp_simple_arxivConfiguration and usage details are provided to help you integrate the server with your MCP client. Use the following setup to enable your client to communicate with the server and perform searches, retrieve paper details, and obtain full-text content in Markdown.
To use the server with a Claude Desktop-like client, configure the MCP connection to start the server as a local stdio process.
{
"mcpServers": {
"simple-arxiv": {
"command": "python",
"args": ["-m", "mcp_simple_arxiv"]
}
}
}