ChEMBL MCP Server

What is ChEMBL MCP Server?

An unofficial Model Context Protocol (MCP) server providing advanced access to the ChEMBL chemical database. It offers 22 specialized tools enabling AI assistants and MCP clients to perform drug discovery research, chemical informatics analysis, and bioactivity investigations through ChEMBL's REST API. Developed by Augmented Nature.

How to use ChEMBL MCP Server?

Install Node.js v16+, clone the repository, run npm install and npm run build, then start with npm start. Add the server to your MCP client configuration (e.g., Claude Desktop) using the path to the compiled build/index.js. Alternatively, run the Docker image with docker run -i chembl-mcp-server.

Key features of ChEMBL MCP Server

• 22 specialized tools for chemical and drug discovery searches
• Compound search, similarity, and substructure retrieval
• Target analysis and UniProt integration
• Bioactivity, assay, and dose-response data access
• Drug development status and clinical trial information
• ADMET analysis, molecular descriptors, and drug-likeness assessment
• Batch processing and external database cross-references

Use cases of ChEMBL MCP Server

• Search for compounds by name or identifier and retrieve detailed chemical properties
• Find biological targets and their associated pathways and mechanisms
• Analyze bioactivity measurements like IC50 or Ki for a target or compound
• Evaluate ADMET properties and drug-likeness for candidate compounds
• Batch-lookup multiple ChEMBL IDs for cross-database comparisons

FAQ from ChEMBL MCP Server

What runtime and dependencies are required?

Node.js v16 or higher, npm or yarn, and the packages @modelcontextprotocol/sdk, axios, and typescript.

Does the server require an API key or authentication?

No. The server uses the public ChEMBL REST API with no authentication. All requests include a User-Agent header ChEMBL-MCP-Server/1.0.0.

Where does the data come from?

All data is retrieved on the fly from the ChEMBL REST API at https://www.ebi.ac.uk/chembl/api/data. No local database is stored.

What transport does the server use?

The server communicates via stdio, as defined by the Model Context Protocol. It is designed to be configured as an MCP client tool.

Are there any usage limits or timeouts?

API requests have a timeout of 30 seconds. Pagination parameters limit (1–1000, default 25) and offset are available on search tools.