Searching PubMed or ClinicalTrials.gov alone is something a browser does. Asking one question and getting trials, published literature and FDA adverse-event data correlated for the same drug is the part that is tedious by hand, because the three sources use different identifiers and different vocabularies. Treat the output as a research starting point rather than a conclusion — the sources are authoritative, the correlation between them is this server's own.
An MCP server that puts three medical research sources behind one set of tools, and then adds tools that query all three together. Results are cached — an hour for clinical trials, six hours for literature — so repeated questions in one session do not re-hit the upstream APIs.
- Clinical trials searched by condition, intervention and page size, and any study read in full by its NCT ID — `ct_search_trials`, `ct_get_study`
- PubMed literature searched for papers on a topic — `pm_search_papers`
- The FDA drug database searched, and adverse events analysed for a drug — `fda_search_drugs`, `fda_adverse_events`
- One drug analysed across all three sources at a depth you choose, returning trials, literature and regulatory data together — `research_comprehensive_analysis`
- A safety picture assembled from trials and FDA data at once, rather than read from each separately — `research_drug_safety_profile`
- The competitive picture for a drug or condition — what else is in trials, from whom — `research_competitive_landscape`
No key is needed to start: all three sources are public. `PUBMED_API_KEY` and `FDA_API_KEY` are optional and raise the rate limits on those two. Node.js 18+. The npm package name is `medical-research-mcp-suite` (1.0.0); install from a clone with `npm install` and `npm run build`, then point your client at `dist/index.js`. `CACHE_TTL` and `MAX_CONCURRENT_REQUESTS` tune how hard it goes at the upstream APIs.
One command — npx -y github:ezhou89/medical-research-mcp-suite