PubChem

This MCP server provides programmatic access to PubChem data via a structured MCP interface. It exposes tools for chemical search, structure analysis, property descriptors, bioassay data, safety/toxicity, and cross-database references, enabling workflows from target discovery to compound evaluation. After starting the server, clients can invoke the predefined tools (for example search_compounds, get_compound_info, search_similar_compounds, get_compound_properties, get_safety_data, batch_compound_lookup, etc.) using MCP messages that specify the tool name and a set of arguments. The server is designed to work in a modular fashion, allowing pipelines that chain compound lookup with property calculations, bioassay retrieval, and safety profiling as part of an end-to-end discovery workflow.

Prerequisites:

• Node.js (14.x or newer) and npm installed on your system
• Git (optional, for cloning the repo)

Install and run:

1. Clone or download the server repository
2. Navigate to the server directory
3. Install dependencies npm install
4. Build the server (if a build step exists in this project) npm run build
5. Start the server npm start

Notes:

• If the project uses a different build or start script (for example, npm run start:prod or node dist/server.js), adjust the commands accordingly.
• Ensure network access to PubChem data sources if not using a fully offline mode.

Tips and common considerations:

• Environment variables: You can tune PORT, PUBCHEM_API_BASE, and LOG_LEVEL to suit deployment needs. Consider setting API rate limit controls if you proxy PubChem requests under heavy load.
• Data sources: The server relies on PubChem REST endpoints; ensure outbound network access and respect PubChem usage guidelines.
• Error handling: If you encounter invalid IDs or malformed tool arguments, the MCP server should return structured error messages. Validate inputs on the client side before sending tool invocations.
• Performance: For batch operations (e.g., batch_compound_lookup), use appropriate max_records limits to prevent large, slow queries. Consider enabling caching if supported by your deployment.
• Versioning: Keep dependencies updated and align MCP tool definitions with the server’s current tool registry to avoid tool-name mismatches.