This is the official remote Model Context Protocol endpoint for the STRING protein-protein interaction database. It is intended for MCP-compatible AI agents and applications.
https://mcp.string-db.org/
Use this URL in an MCP client that supports remote servers over Streamable HTTP.
MCP, the Model Context Protocol, is a standard way for AI applications to connect to external tools and data sources. The STRING MCP server lets an AI agent access STRING data, for example to resolve protein identifiers, retrieve interaction networks, inspect evidence, run enrichment analysis, look up homologs, or query proteins by biological function or disease association.
Open the settings or connector configuration in an MCP-compatible client and add a remote MCP server. The exact interface depends on the client. The fields are typically similar to:
Name: STRING database
URL: https://mcp.string-db.org/Some clients may call this a custom connector, remote MCP server, HTTP MCP server, or simply an MCP server.
After connecting, ask your AI assistant biological questions that require STRING data, for example:
Find interaction partners of TP53 in human.
Run functional enrichment for these proteins: TP53, MDM2, CDKN1A, ATM.
Map these gene names to STRING identifiers.
Find homologs of human EGFR in mouse.STRING also provides a live chat interface where you can try the MCP server with STRING data directly. This is a convenient way to test the available tools before configuring your own agent.
Go to string-db.org/cgi/chat.
The hosted endpoint above is the usual way to use the STRING MCP server. The open source server code can also be run locally for development, extension, testing, or reproducible workflows.
Source code: github.com/meringlab/string-mcp
git clone https://github.com/meringlab/string-mcp.git
cd string-mcp
pip install -r requirements.txt
python server.pyFor non-agentic programmatic access to STRING, use the regular STRING API instead.