entrez

This MCP server exposes NCBI Entrez APIs (E-utilities), PubChem PUG, and PMC services via the MCP protocol. It is designed to work out of the box with no configuration required, but you can optionally supply an NCBI API key to unlock higher rate limits. Once running, you can interact with the MCP endpoints using tool IDs in your MCP-compatible clients, such as Claude Desktop or the Cloudflare AI Playground, to perform queries across NCBI resources, fetch metadata, or automate data retrieval tasks. If you’re using the Cloudflare AI Playground, point the playground’s remote MCP URL to your server URL and start sending requests to the available Entrez tools.

You can test your setup locally by invoking the provided test utilities. Without an API key, you’ll be limited to the default rate, but with an API key you’ll be able to push closer to the elevated limit. The server is designed to respect NCBI rate limits and provides a straightforward way to run common workflows like E-utilities queries, PubChem lookups, and PMC document access through MCP tools.

Prerequisites:

• Node.js (LTS) and npm installed on your machine
• A working internet connection to install dependencies

1. Install dependencies

   • git clone <this-repo>
   • cd entrez-mcp-server
   • npm install

2. Run the server (default, out-of-the-box configuration)

   • npm start

3. Optional: supply an NCBI API key for higher rate limits

   • In your environment, set: export NCBI_API_KEY="your_key_here"
   • Then restart the server: npm start

4. Verify the server is running

   • Open a browser or use curl to access the MCP endpoint exposed by the server (e.g., http://localhost:8787/mcp)

5. Optional: connect clients

   • Cloudflare AI Playground: use your server URL as described in the README to interact with MCP tools
   • Claude Desktop: configure the mcpServers section to point to your local or remote MCP server as shown in the README

Tips and common considerations:

• If you don’t provide an API key, expect rate limits around 3 requests/second; providing an API key increases this to about 10 requests/second.
• To connect Cloudflare AI Playground or Claude Desktop, ensure your MCP server is publicly reachable if you’re not running locally, or configure port forwarding / reverse proxies as needed.
• For Cloudflare Code Mode users: prefer underscore tool IDs (e.g., tool_entrez_query) to avoid JavaScript parsing issues; the hyphenated aliases remain for compatibility but may require bracket notation when invoked programmatically.
• If you encounter rate-limit warnings, consider batching requests or adding a proper NCBI API key to maximize throughput.
• No extra configuration is required for basic usage; the server is designed to be plug-and-play with the default setup.