NCBI MCP Server

NCBI MCP Server connects Claude to the National Center for Biotechnology Information's databases, primarily PubMed and the MeSH taxonomy. If you're writing grant proposals, literature reviews, or building tools for life sciences research, this is a direct line to the world's largest biomedical literature archive without leaving your editor. What makes this worth installing is the caching layer and batch processing. PubMed searches can be slow and repetitive, especially when you're iterating on search terms or pulling related articles. The server caches results locally, so you're not hammering NCBI's API every time you refine a query. Batch operations let you pull metadata for dozens of papers in one go, which is genuinely useful if you're building reference lists or comparing citation networks. The MeSH integration is solid: you can browse the controlled vocabulary tree, find term definitions, and map free-text concepts to official headings. I'd reach for this when building a systematic review pipeline or when I need Claude to suggest related papers based on a starting set. The quirks are real but manageable. You need an NCBI API key, which is free but requires registration. Without it, you're rate-limited to three requests per second, which will slow down batch jobs. The install uses Poetry, so if you're not already in that ecosystem, expect a few extra minutes of setup. The documentation assumes you know what MeSH is and why you'd want it; if you're not in biomedical research, the taxonomy features will feel opaque. Related article discovery works well for recent papers but can return sparse results for older or niche publications. This isn't for casual use. If you're not regularly searching PubMed or don't know what a MeSH heading is, you won't get much from it. But for computational biologists, medical writers, or anyone building research automation, it's a clean, focused tool that does one thing well. The caching alone saves enough time to justify the setup.