genomoncology/biomcp

curl -fsSL https://biomcp.org/install.sh | bash
uv tool install biomcp-cli  (or pip install biomcp-cli)
brew tap genomoncology/biomcp && brew install biomcp
Docker: docker run --rm ghcr.io/genomoncology/biomcp serve

Usage

• •Register the stdio MCP server: {"mcpServers": {"biomcp": {"command": "biomcp", "args": ["serve"]}}}
• •Claude Code plugin: /plugin marketplace add genomoncology/biomcp then /plugin install biomcp@biomcp; install skills with `biomcp skill install ~/.claude --force`
• •Codex: codex mcp add biomcp -- biomcp serve; remote HTTP: biomcp serve-http --host 127.0.0.1 --port 8080 (clients connect to /mcp)
• •Core grammar: biomcp search all --gene BRAF --disease melanoma, biomcp get gene BRAF pathways hpa, biomcp variant articles "BRAF V600E", biomcp enrich BRAF,KRAS,NRAS

Key features

• ✓Unified command grammar over ~30 sources: search/get/helpers/enrich/batch with sections and progressive disclosure per entity (gene, variant, article, trial, diagnostic, drug, disease, pathway, protein, adverse-event, pgx)
• ✓Literature search fans out across PubTator3 and Europe PMC with PMID/PMCID/DOI dedup and optional Semantic Scholar leg
• ✓Cross-entity pivots (variant→trials, gene→pathways, drug→adverse-events) and article paper-trail (citations/references/recommendations/entities)
• ✓Local study analytics on downloaded cBioPortal-style datasets: study query/cohort/survival/compare/co-occurrence with terminal, SVG, and PNG charts
• ✓g:Profiler gene-set enrichment and batch (up to 10 focused get calls); JSON responses expose _meta.next_commands and section provenance

Best for

Biomedical researchers and clinicians who want one command grammar to query and pivot across major public biomedical databases from a CLI or an AI agent.

Caveats

• ⚠PyPI warning: install biomcp-cli, NOT biomcp (the biomcp PyPI package is unrelated)
• ⚠Most commands work without credentials; optional keys (NCBI_API_KEY, S2_API_KEY, OPENFDA_API_KEY, NCI_API_KEY, ONCOKB_TOKEN, ALPHAGENOME_API_KEY) improve rate limits or unlock sections
• ⚠KEGG distinguishes academic and non-academic use; COSMIC is kept indirect-only due to licensing incompatibility
• ⚠Article references/recommendations can be empty for paywalled papers (publisher elision in Semantic Scholar coverage)
• ⚠Rate limiting is process-local - use one shared serve-http endpoint for many concurrent workers