About Biomcp
BioMCP is a single binary command line tool and MCP server that gives researchers, clinicians and AI agents one consistent way to query biomedical databases. Rather than juggling separate websites, authentication schemes and APIs, users search, open detailed records and pivot between related entities with one command grammar, and the output stays compact enough for an agent's context window.
It covers genes, variants, articles, authors, clinical trials, diagnostics, drugs, diseases, pathways and proteins, drawing on sources such as MyGene.info, UniProt, Reactome, ClinVar, gnomAD, CIViC, OncoKB, cBioPortal, PubMed, PubTator3, Europe PMC, Semantic Scholar, ClinicalTrials.gov, the NCI trials API and ChEMBL. A discover command resolves aliases, brand names and symptoms to the right entity, article commands follow citations and references, enrich runs gene set enrichment through g:Profiler, and batch runs up to ten lookups at once. It can also analyse downloaded cBioPortal style study datasets locally, with cohort, survival and co occurrence workflows and chart output. Installable skills add guided investigation workflows for agents, and setup helpers configure Claude Code, Claude Desktop, Cursor, Codex, Cline and VS Code. It runs over standard input or as a remote HTTP server.
BioMCP is free and MIT licensed, installable with a shell script or from PyPI. It queries upstream providers on demand rather than mirroring their data, so each source's own terms govern reuse, and some features require free registration or API keys from those providers.