home / mcp / paper search mcp server

Paper Search MCP Server

A MCP for searching and downloading academic papers from multiple sources like arXiv, PubMed, bioRxiv, etc.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "adamamer20-paper-search-mcp-openai": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/path/to/your/paper-search-mcp",
        "-m",
        "paper_search_mcp.server"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

You can run a Python-based MCP server that lets you search and download academic papers from multiple sources, enabling seamless integration with language models and automated workflows. This server exposes standardized search and fetch capabilities so your AI agents can discover relevant papers and obtain their PDFs efficiently.

How to use

To use the paper search MCP server, you run it locally and connect your MCP client to the provided runtime. You can then perform searches across supported sources (such as arXiv, PubMed, bioRxiv, and more) and fetch downloadable PDFs for the results. The quick start shows how to launch the server and configure your client-software to point at it. If you are using Claude Desktop, you will typically register the MCP server in your Claude configuration so that search and fetch tools are available within your conversational workflows.

How to install

Prerequisites: you need Python-based tooling installed and access to the MCP runtime used by your environment. You will also need the ability to run local commands via your MCP client when connecting to this server.

# Note: This snippet shows the recommended local runtime invocation from the quick start configuration.
# It uses the standard MCP runtime command to launch the server in your environment.
uv run --directory /path/to/your/paper-search-mcp -m paper_search_mcp.server

Configuration examples

The server can run as a local process and be reached by your MCP client through the runtime. If you are configuring a client like Claude Desktop, point it to the local process using the provided command and directory path. The optional semantic scholar API key can be supplied to enable enhanced features.

Notes on environment

Optional environment variables may be used to enable extra data sources or features. For example, you can provide a Semantic Scholar API key to enable deeper integrations with that platform.

Available tools

search_arxiv

Search papers from arXiv and other sources; returns a list of results with metadata.

download_arxiv

Download the PDF for a selected arXiv paper by its identifier.

search

General search tool across integrated sources returning standardized results.

fetch

Fetch the full Paper data, including metadata and PDF link, for a given result.

Paper

Structured data class representing a paper with fields like title, authors, source, id, and pdf URL.