paper-search

paper-search-mcp is a Python-based MCP server that lets you search for academic papers across multiple sources (such as arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, IACR ePrint Archive, and Semantic Scholar) and download PDFs when needed. It exposes tools like search_arxiv for querying arXiv, download_arxiv for retrieving PDFs, and similar tools for other platforms, all designed to integrate with MCP clients used by large language models. When deployed via MCP, you can request paper metadata, obtain structured results, and optionally fetch full-text PDFs to augment the context available to your LLM workflows. The server is built to work with Claude Desktop and other MCP-enabled clients, allowing you to streamline literature gathering and retrieval as part of your AI-assisted research pipelines.

Prerequisites:

• Python 3.10+ (per project requirements)
• Git
• Optional: uv (for development/usage) or pip/poetry as preferred package manager

Installation steps:

1. Install uv (if not already installed) and set up a Python environment

• Install uv if needed:

  curl -LsSf https://astral.sh/uv/install.sh | sh

• Clone the repository (if you have not yet):

  git clone https://github.com/openags/paper-search-mcp.git cd paper-search-mcp

2. Install the package in editable mode (development workflow) or install from PyPI as appropriate

• Development install (editable):

  uv add -e .

• Optional: install development dependencies (pytest, flake8, etc.):

  uv add pytest flake8

3. Run the MCP server using uv (example):

uv run --directory /path/to/your/paper-search-mcp -m paper_search_mcp.server

4. If you want to configure Claude Desktop, add the provided mcpServers configuration in Claude’s config file (see README example).

Notes and tips:

• The server supports multiple sources; ensure you have API keys where required (e.g., SEMANTIC_SCHOLAR_API_KEY) for enhanced features. The README shows SEMANTIC_SCHOLAR_API_KEY as an optional environment variable.
• Adjust the --directory path to point to your local paper-search-mcp installation when embedding in Claude Desktop or other MCP clients.
• If you modify code or add new platforms, update the academic_platforms module accordingly and run tests to verify MCP tool signatures (e.g., search_arxiv, download_arxiv).
• Ensure your Python environment and dependencies are kept up to date to avoid compatibility issues with the MCP SDK.
• The MCP server is designed to be easily extensible; you can add new platforms by extending the academic_platforms module and exposing corresponding tools via the MCP API.