u9401066/pubmed-search-mcp

View on GitHub ↗

MCP SDK v2 server for PubMed and multi-source biomedical research, with resilient search, full text, citations, and tenant-safe deployment.

25 ★8 forksPythonUpdated 18d ago

What you need to know

A domain-driven MCP literature research assistant with 45 tools for searching PubMed, Europe PMC, CORE, OpenAlex and preprint servers, with vocabulary translation (ICD to MeSH), full-text retrieval, citation trees and exports.

Install

uvx pubmed-search-mcp
pip install pubmed-search-mcp

Usage

  • unified_search(q) searches multiple sources with dedup and ranking
  • Use parse_pico() for structured clinical PICO searches

Key features

  • 45 MCP tools over multiple academic sources
  • Auto ICD to MeSH conversion and ESpell correction
  • OA figure extraction and multi-source fulltext
  • Citation trees, research chronicle and context graph
  • RIS/BibTeX/CSV/MEDLINE export
  • Multi-agent service mode with per-tenant isolation

Best for

Academic literature search, clinical research and citation analysis

Caveats

  • NCBI_EMAIL is required by NCBI API policy
  • Cache and sessions are tied to the default tenant in local mode
  • Not supported yet for the fulltext browser-fetch path without extra setup
Platforms: macOS · Windows · LinuxClients: Claude Desktop · Claude Code · VS Code · Cursor · Zed · Cline · OpenClaw

Reviewed 2026-08-11

Topics

biomedical-researchcitation-analysisentrezeurope-pmcevidence-based-medicinefulltextliterature-searchmcpmcp-servermodel-context-protocolmulti-tenantncbiopen-accessopenalexpicopubmedpythonsemantic-scholarstreamable-httpsystematic-review
Stars
25★
Forks
8
Language
Python
License
NOASSERTION
Created
2025-12-02
Last push
2026-08-17