Simple PubMed MCP

Simple PubMed MCP does exactly what the name suggests: it searches PubMed's medical literature database and returns either full-text articles (when open-access) or abstracts. It's built on the Entrez API, which is the same backend that powers the PubMed website, so you're getting the real deal. I'd reach for this when I'm writing anything that needs medical citations or when I'm trying to understand a clinical topic without leaving my editor. The workflow is straightforward: ask Claude to search PubMed for recent studies on, say, metformin and insulin resistance, and you get back structured results with titles, authors, abstracts, and DOIs. If the article is open-access, you get the full text. Otherwise, you get the abstract, which is usually enough to decide if you want to track down the paper. The main quirk is that you need to provide an email address via the PUBMED_EMAIL environment variable. This isn't for authentication, it's a politeness requirement from NCBI so they can contact you if your queries are hammering their servers. It's a sensible policy, but it does mean one extra setup step. What it's good for: clinical research, evidence-based writing, drug interaction checks, or any time you need to cite peer-reviewed medical literature. If you're writing health content or doing biomedical research, this is a genuinely useful tool. The full-text access for open-access articles is particularly handy, it saves you the trip to Sci-Hub or your institutional login. What it's not: a replacement for a proper reference manager, and it won't give you full-text access to paywalled journals unless they're open-access. The search is also only as good as your query, PubMed's syntax can be fiddly if you're trying to do complex Boolean searches. Who shouldn't bother: if you're not working in health, medicine, or life sciences, you won't need this. It's a specialist tool for a specialist database. Developers building general-purpose AI agents can skip it unless their users are clinicians or researchers.