Tools

Ten tools for querying PubChem's chemical information database:

pubchem_search_compounds

Search PubChem for chemical compounds across five search modes.

• Identifier lookup — resolve compound names, SMILES, or InChIKeys to CIDs (batch up to 25)
• Formula search — find compounds by molecular formula in Hill notation
• Substructure/superstructure — find compounds containing or contained within a query structure
• 2D similarity — find structurally similar compounds by Tanimoto similarity (configurable threshold)
• Caps at 200 CIDs per page; offset pages further, to a ceiling of 10,000. Identifier lookups page over the set already resolved; formula and structure searches widen their bounded upstream request to reach a page, so deep pages cost more upstream
• Optionally hydrate results with properties to avoid a follow-up details call

pubchem_get_compound_details

Get detailed compound information by CID.

• Batches up to 100 CIDs in a single request
• 27 available properties: molecular weight, SMILES, InChIKey, XLogP, TPSA, complexity, stereo counts, and more
• Optionally includes textual descriptions (pharmacology, mechanism, therapeutic use) from PUG View — fetched for the first 10 CIDs of a batch, with the skipped CIDs named in the response
• Optionally includes known synonyms (trade names, systematic names, registry numbers)
• Synonyms and descriptions are paged: synonymOffset and descriptionOffset window every compound in the batch at the same position, reaching the entries past a page
• Optionally computes drug-likeness assessment (Lipinski Rule of Five + Veber rules) from fetched properties
• Optionally fetches pharmacological classification (FDA classes, mechanisms of action, MeSH classes, ATC codes)

pubchem_get_bioactivity

Get a compound's bioactivity profile from PubChem BioAssay.

• Returns assay outcomes (Active/Inactive/Inconclusive), target info (protein accessions, NCBI Gene IDs), and quantitative values (IC50, EC50, Ki)
• Filter by outcome and/or a specific molecular target (NCBI Gene ID or protein accession)
• Caps at 100 results per page; offset reaches the rest (well-studied compounds may have thousands)

pubchem_get_summary

Get descriptive summaries for four PubChem entity types.

• Assays (AID), genes (Gene ID), proteins (UniProt accession), taxonomy (Tax ID)
• Up to 10 entities per call
• Type-specific field extraction for clean, structured output

pubchem_get_compound_interactions

Get a compound's interaction data by CID.

• Drug-drug interactions (DrugBank), drug-food interactions, and chemical-target binding/activity (BindingDB, ChEMBL, and others)
• Select which interaction kinds to fetch and cap entries per kind
• Paged per kind: each reports its source-record total and its own nextOffset, and offset reaches the records past a page
• Each entry carries its originating source — coverage is richest for approved drugs

pubchem_get_compound_3d_structure

Get a compound's default 3D conformer by CID.

• format="json" returns parsed atoms (element + x/y/z) and bonds for direct reasoning; format="sdf" returns raw V2000 SDF for passthrough to docking or rendering
• maxAtoms/maxBonds bound the atom/bond preview and includeRawSdf opts into a large raw SDF past the safe line cap; atomCount/bondCount always report the totals and any capping is disclosed
• Optionally lists alternate conformer IDs
• Returns a typed not-found when PubChem has no computed 3D coordinates (large molecules, mixtures, some salts)

Resources

Compound and assay records are also exposed as URI-templated MCP resources, backed by the same client methods as the tools:

Features

Built on @cyanheads/mcp-ts-core:

• Declarative tool definitions — single file per tool, framework handles registration and validation
• Unified error handling across all tools
• Pluggable auth (none, jwt, oauth)
• Swappable storage backends: in-memory, filesystem, Supabase, Cloudflare KV/R2/D1
• Structured logging with optional OpenTelemetry tracing
• Runs locally (stdio/HTTP) or containerized via Docker

PubChem-specific:

• Rate-limited client for PUG REST and PUG View APIs (5 req/s with automatic queuing)
• Retry with exponential backoff on 5xx errors and network failures
• All tools are read-only and idempotent — no API keys required

Getting Started

Public Hosted Instance

A public instance is available at https://pubchem.caseyjhand.com/mcp — no installation required. Point any MCP client at it via Streamable HTTP:

{
  "mcpServers": {
    "pubchem-mcp-server": {
      "type": "streamable-http",
      "url": "https://pubchem.caseyjhand.com/mcp"
    }
  }
}

Self-Hosted / Local

Add to your MCP client config (e.g., claude_desktop_config.json):

{
  "mcpServers": {
    "pubchem-mcp-server": {
      "type": "stdio",
      "command": "bunx",
      "args": ["@cyanheads/pubchem-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio"
      }
    }
  }
}

Prerequisites

• Bun v1.3.0 or higher (or Node.js v24+)

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/pubchem-mcp-server.git

1. Navigate into the directory:

cd pubchem-mcp-server

1. Install dependencies:

bun install

Configuration

No API keys are required — PubChem's API is freely accessible.

Running the Server

Local Development

• Build and run:

  sh

  Copy

  bun run rebuild
  bun run start:stdio   # or start:http
  

• Run checks and tests:

  sh

  Copy

  bun run devcheck     # Lints, formats, type-checks
  bun run test         # Runs test suite
  

Docker

docker build -t pubchem-mcp-server .
docker run -p 3010:3010 pubchem-mcp-server

Project Structure

Development Guide

See CLAUDE.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for domain-specific logging
• Register new tools in the index.ts barrel file

Contributing

Issues and pull requests are welcome. Run checks before submitting:

bun run devcheck
bun run test

License

Apache-2.0 — see LICENSE for details.