PubMed MCP Server

A Model Context Protocol (MCP) server that provides access to PubMed's E-utilities API for searching and downloading scientific articles. This server enables LLM applications to search PubMed's vast database of biomedical literature and retrieve article metadata, abstracts, and full content.

- Article Search: Search PubMed database with flexible query terms
- Article Download: Retrieve full article metadata, abstracts, and available content
- Batch Operations: Download multiple articles in a single request
- Article Summaries: Get document summaries with metadata
- Multiple Formats: Support for XML, JSON, and text output formats
- Rate Limiting: Automatic rate limiting to respect PubMed API limits
- Error Handling: Robust error handling for API failures
- Clone or downloadthis repository
- Run the setup script:

./setup.sh

- Clone or downloadthis repository
- Create and activate virtual environment:

python3 -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate

cp .env.example .env # Edit .env file with your NCBI API key and email

Create a.envfile with the following optional configuration:

- NCBI_API_KEY: Your NCBI API key (increases rate limit from 3 to 10 requests/second)
- NCBI_EMAIL: Your email address (recommended by NCBI for API usage tracking)

Get your free NCBI API key at:https://www.ncbi.nlm.nih.gov/account/settings/
-

source venv/bin/activate # On Windows: venv\Scripts\activate

deactivate

{ "query": "CRISPR gene editing", "max_results": 10, "sort": "pub_date" }

- "COVID-19 AND vaccine"- Articles about COVID-19 vaccines
- "machine learning AND healthcare"- ML in healthcare
- "CRISPR[Title]"- CRISPR in article titles only
- "Nature[Journal]"- Articles from Nature journal
- "2023[PDAT]"- Articles published in 2023
- "Smith J[Author]"- Articles by author "Smith J"

- [Title]- Search in title only
- [Author]- Search by author
- [Journal]- Search by journal name
- [PDAT]- Search by publication date
- [MeSH]- Search MeSH terms

If you configured your API key in the.envfile during installation:

{ "mcpServers": { "pubmed": { "command": "/path/to/pubmed-mcp/venv/bin/python", "args": ["/path/to/pubmed-mcp/server.py"] } } }

Alternatively, you can specify the API key directly in the Claude Desktop configuration:

{ "mcpServers": { "pubmed": { "command": "/path/to/pubmed-mcp/venv/bin/python", "args": ["/path/to/pubmed-mcp/server.py"], "env": { "NCBI_API_KEY": "your_api_key_here", "NCBI_EMAIL": "your_email@example.com" } } } }

Recommendation: Use Option 1 (.env file) for better security and easier management.

Note: Make sure to use the full path to the Python executable in the virtual environment (venv/bin/python) to ensure the correct dependencies are available.

- Without API key: 3 requests per second
- With API key: 10 requests per second
- Batch size limit: 50 articles per batch request

The server provides comprehensive error handling:

- Invalid PMIDs are automatically cleaned (non-numeric characters removed)
- Empty queries return descriptive errors
- API failures are caught and reported
- Rate limiting prevents API abuse

pubmed-mcp/ ├── server.py # Main MCP server implementation ├── pubmed_client.py # PubMed API client wrapper ├── requirements.txt # Python dependencies ├── setup.sh # Automated setup script ├── .gitignore # Git ignore file ├── README.md # This file ├── .env.example # Environment variables template └── venv/ # Virtual environment (created by setup)

- mcp[cli]- MCP Python SDK
- requests- HTTP client for PubMed API
- python-dotenv- Environment variables
- typing-extensions- Type hints support

This project is open source. Please check PubMed's terms of service for API usage guidelines.

For issues with this MCP server, please check:
- Your API key and email configuration
- Network connectivity to NCBI servers
- Rate limiting compliance
- Valid PMID formats

For PubMed API documentation, visit:https://www.ncbi.nlm.nih.gov/books/NBK25500/

Search and access clinical trial data from ClinicalTrials.gov.