MCP Simple PubMed

To automatically install Simple PubMed for Claude Desktop, use the Smithery CLI with the following command:

npx -y @smithery/cli install mcp-simple-pubmed --client claude

To use the server with Claude Desktop, add the following configuration to your claude_desktop_config.json file. Replace the placeholders with your actual email and API key.

// macOS Configuration
{
  "mcpServers": {
    "simple-pubmed": {
      "command": "python",
      "args": ["-m", "mcp_simple_pubmed"],
      "env": {
        "PUBMED_EMAIL": "your-email@example.com",
        "PUBMED_API_KEY": "your-api-key" 
      }
    }
  }
}

You can install the server manually using pip:

pip install mcp-simple-pubmed

On Windows, you must provide the full path to your Python executable in the command field of your claude_desktop_config.json.

{
  "mcpServers": {
    "simple-pubmed": {
      "command": "C:\\Users\\YOUR_USERNAME\\AppData\\Local\\Programs\\Python\\Python311\\python.exe",
      "args": [
        "-m",
        "mcp_simple_pubmed"
      ],
      "env": {
        "PUBMED_EMAIL": "your-email@example.com",
        "PUBMED_API_KEY": "your-api-key" 
      }
    }
  }
}

The server requires specific environment variables to interact with the NCBI Entrez API.

• PUBMED_EMAIL: Required. Your email address (required by NCBI).
• PUBMED_API_KEY: Optional. An API key to increase rate limits from 3 requests/second to 10 requests/second.

If you encounter [SSL: CERTIFICATE_VERIFY_FAILED] errors on macOS, you must install the Python certificate bundle. Run the following command in your terminal, replacing 3.13 with your specific Python version:

/Applications/Python\ 3.13/Install\ Certificates.command

Note: You must restart Claude Desktop for this change to take effect.

The server requires specific environment variables to initialize the PubMedClient and FullTextClient.

Required Environment Variables:

• PUBMED_EMAIL: Your email address (required for PubMed API usage).

Optional Environment Variables:

• PUBMED_TOOL: The name of the tool (defaults to mcp-simple-pubmed).
• PUBMED_API_KEY: Your PubMed API key.

The server provides the following tools for interacting with PubMed:

• Search PubMed: Search using keywords, MeSH terms, author names, date ranges, and Boolean operators.
• Get Full Text: Downloads full text for open access articles in PubMed Central. Note that the returned full text is an XML-ized version, which provides structural information useful for AI models.
• Access Abstracts: Retrieves article abstracts and metadata via resource URIs.

As of v0.1.14, the server includes prompts to help construct optimized PubMed queries:

• Systematic Review Search: Generates comprehensive strategies using MeSH terms, synonyms, and date filters.
• PICO Search: Builds clinical question searches using the Population, Intervention, Comparison, Outcome framework.
• Author Search: Finds all publications by a specific author with proper name formatting.

To start the MCP server, call the main() function. This function initializes the asynchronous event loop and runs the server's main entry point defined in the package's server module.

from mcp_simple_pubmed import main

if __name__ == "__main__":
    main()

Search PubMed for medical and life sciences research articles using keyword or field-specific queries.

Search Features:

• Simple keyword search: "covid vaccine"
• Field-specific search:
  • Title: [Title]
  • Author: [Author]
  • MeSH terms: [MeSH Terms]
  • Journal: [Journal]
• Date ranges: Use "2020:2024"[Date - Publication]
• Boolean operators: Combine terms with AND, OR, NOT
• Exact phrases: Use quotation marks for multi-word terms.

Returns: A JSON string containing paper titles, authors, publication details, abstract previews, links to full text, DOIs, and keywords/MeSH terms.

Constraints: max_results is constrained between 1 and 50.

# Example usage via tool call
await search_pubmed(query='"breast cancer"[Title] AND "2023"[Date - Publication]', max_results=10)