I'd like to share an open-source tool I developed for integrating KEGG into AI-assisted bioinformatics workflows.

In our lab we work with bacterial and fungal genomics and metabolomics, and KEGG is a constant reference — pathway annotation, compound lookups, orthology mapping, metabolic model contextualization. I built this tool to stop rewriting the same REST wrappers and let AI assistants query KEGG directly during analysis sessions.

kegg-mcp-server-python (GitHub) is a Python implementation of a Model Context Protocol (MCP) server for the KEGG database. MCP is an open standard that lets AI assistants (Claude, Cursor, etc.) call external tools in a structured way — giving your AI client direct, programmatic access to KEGG without prompt engineering around raw text.

It is a Python port and extension of the original JavaScript implementation by Augmented-Nature.

What it provides

33 tools covering pathways, genes, compounds, reactions, enzymes, diseases, drugs, modules, orthology (KO), glycans, and BRITE hierarchies Cross-database operations: batch entry lookup (up to 50 entries), identifier conversion between KEGG and UniProt/NCBI/ChEBI/PubChem, and related-entry discovery 3 guided analysis prompts for pathway enrichment, drug target investigation, and cross-species metabolic pathway comparison All responses are structured Pydantic JSON — not raw flat-file text — which makes downstream parsing reliable No API key required — uses the free KEGG REST API at rest.kegg.jp

Installation

# Run directly without installing
uvx kegg-mcp-server

# Or install via pip
pip install kegg-mcp-server
Compatible with Claude Desktop, Claude Code, Cursor, and any MCP-compliant client.

Feedback and contributions are welcome, especially from people working with non-model organisms where KEGG coverage can be patchy. Happy to answer questions about the implementation or specific use cases.