pubmed-mcp
Related Servers
Alternatives to pubmed-mcp
No user-submitted related servers found.
Related Servers
- AlicenseNot gradedqualityBmaintenanceEnables searching and retrieving biomedical literature from PubMed via the NCBI E-utilities API.225 npm1MIT
- FlicenseNot gradedqualityCmaintenanceEnables searching PubMed's biomedical literature database and retrieving article metadata, abstracts, and full content through the E-utilities API. Supports advanced queries, batch operations, and multiple output formats with automatic rate limiting.2-
- AlicenseNot gradedqualityCmaintenanceEnables PubMed literature search, metadata retrieval, BibTeX export, and evidence table generation for biomedical research agents.2MIT
- AlicenseBqualityDmaintenanceEnables searching and retrieving scientific articles from PubMed using NCBI E-utilities API with features like rate limiting and caching.338 npmMIT
- AlicenseAqualityCmaintenanceProvides structured PubMed literature data for LLM agents, supporting search, caching, and open-access full-text downloads via the MCP protocol.551 npm11Apache 2.0
- AlicenseAqualityDmaintenanceExposes NCBI PubMed as MCP tools for searching literature, fetching abstracts, exploring citation graphs, and finding author publications without requiring an API key.4MIT
TDQS
Scored across 10 tools
pubmed_search and search_pubmed are functionally near-identical (both run NCBI esearch and return PMIDs/query translation), and pubmed_fetch, fetch_abstracts, and pubmed_batch_fetch all wrap efetch with only minor output differences. check_access vs search_with_access also overlap, leaving several boundaries unclear despite good descriptions.
Names mix a pubmed_ prefix (pubmed_search, pubmed_fetch, pubmed_get, pubmed_batch_fetch, pubmed_database_info) with bare verb_noun forms (fetch_abstracts, check_access, download_pdfs) and an unconventional reversed form (search_pubmed). The search pair differing only by word order is especially confusing.
Ten tools is a reasonable scope for a PubMed retrieval server, but several tools are redundant variants of the same underlying operation rather than distinct capabilities. The count itself is fine; the duplication is the issue.
The surface covers the core retrieval lifecycle well: search, single/multi/batch fetch, normalized or raw output, access classification, and PDF download. Minor gaps exist (no citation/related-article or export formats like RIS/BibTeX), which agents can work around.