One Rust binary replaces a dozen biomedical APIs

What it does

BioMCP is a Rust binary that acts as a thick, well-organized glue layer between roughly a dozen public biomedical APIs—PubMed, Europe PMC, ClinVar, MyGene.info, ClinicalTrials.gov, and others. It exposes a single command grammar to search, fetch, and pivot across genes, variants, drugs, diseases, trials, and articles without forcing you to relearn each upstream identifier scheme. It can also run as a local Model Context Protocol server, letting LLM agents query the same federated sources through a standardized interface.

The interesting bit

The cross-entity pivoting is the real convenience trick: a result about a variant can jump straight to related trials, drugs, or articles using the same identifier, rather than manually re-querying a new API with a new ID format. The CLI also supports progressive disclosure, where get responses expose selectable sections and JSON metadata that suggests the next likely follow-up commands.

Key highlights

• Federated literature search deduplicates PMIDs, PMCIDs, and DOIs across PubTator3, Europe PMC, and an optional Semantic Scholar leg.
• Thirteen entity types are supported, from genes and proteins to adverse events and pharmacogenomics, each drawing from multiple specialized databases.
• Local study analytics with terminal, SVG, and PNG charts for downloaded cBioPortal-style datasets.
• Most commands work without API keys; optional tokens unlock premium sources like OncoKB or higher rate limits.
• Ships with a suggest playbook router that turns natural-language questions into starter commands.

Caveats

• OncoKB therapy evidence, scored DisGeNET associations, and some Semantic Scholar features require optional API tokens; without them, those legs are unavailable or rate-limited.
• Article references and recommendations can return empty results for paywalled papers due to upstream gaps in Semantic Scholar coverage.
• The Claude Desktop extension currently surfaces only three optional settings, so advanced CLI-only environment variables remain manual.

Verdict

Worth a look if you are a researcher, clinician, or agent builder tired of maintaining separate API clients for every public biomedical database. Skip it if you need a primary data repository or deep analytics beyond evidence retrieval and cohort summaries.