Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From pubmed
This skill should be used when the user asks to "search PubMed", "find PubMed articles about", "look up biomedical literature", "search for papers on", "find citations for", "get article abstracts", "find related papers", "format citations in BibTeX", "format citations in APA", "plan a literature review", "generate a research plan", "visualize publication trends", "show a chart of publications", "graph publication counts", or mentions PubMed, NCBI, biomedical literature, PMIDs, MeSH terms, or scientific article retrieval. Provides expert guidance on using the five PubMed MCP server tools for biomedical research workflows.
npx claudepluginhub thorrak/pubmed_pluginHow this skill is triggered — by the user, by Claude, or both
Slash command
/pubmed:pubmed-researchThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill for biomedical literature search and retrieval using the PubMed MCP server. The server exposes five tools via `mcp__plugin_pubmed_pubmed__*` and handles all NCBI API communication.
Constructs advanced PubMed queries with Boolean/MeSH operators and E-utilities API for biomedical literature retrieval. Use for systematic reviews, citation management, or automated literature monitoring.
Provides programmatic access to PubMed via NCBI E-utilities REST API for Boolean/MeSH queries, ESearch/EFetch endpoints, batch processing, citation matching, and biomedical literature pipelines.
Provides direct REST API access to PubMed via NCBI E-utilities for advanced Boolean/MeSH queries, batch processing, and citation management. Use for biomedical literature searches, systematic reviews, or custom API workflows.
Share bugs, ideas, or general feedback.
Use this skill for biomedical literature search and retrieval using the PubMed MCP server. The server exposes five tools via mcp__plugin_pubmed_pubmed__* and handles all NCBI API communication.
| Tool | Purpose | Key Input |
|---|---|---|
pubmed_search_articles | Search PubMed by query | queryTerm (required) |
pubmed_fetch_contents | Retrieve full metadata for known articles | pmids or queryKey+webEnv |
pubmed_article_connections | Find related articles and format citations | sourcePmid (required) |
pubmed_research_agent | Generate structured research plans | project_title_suggestion, primary_research_goal, research_keywords |
pubmed_generate_chart | Create PNG charts from publication data | chartType, dataValues, xField, yField |
Start with search when the user does not have PMIDs:
Call pubmed_search_articles first to get PMIDs and brief summaries. Pass those PMIDs to pubmed_fetch_contents for full abstracts, authors, and DOIs. This two-step pattern is the most common workflow.
Use fetch directly when the user provides PMIDs:
Skip the search step. Call pubmed_fetch_contents with the known PMIDs. Set detailLevel to full for abstracts and author affiliations, or summary for a lighter metadata response.
Use article connections for literature mapping:
Call pubmed_article_connections to find citing papers (cited_by), similar articles (similar), or references (references). This tool also formats citations—pass citationStyles: ["APA", "BibTeX"] to get formatted output. Combine with a search to expand a bibliography from a single seed article.
Use the research agent for structured planning:
Call pubmed_research_agent when the user wants a systematic literature review strategy, not just individual results. It returns a JSON research plan with recommended search strings, Boolean operators, and MeSH term suggestions. Use the output to drive subsequent pubmed_search_articles calls.
Use chart generation for trend visualization:
Call pubmed_generate_chart when the user wants a visual of publication trends (e.g., articles per year), comparison across topics, or distribution by journal. Feed it data derived from prior search results.
Simple keyword queries work for most cases:
"CRISPR cancer therapy" — natural language"Smith J[Author] AND Nature[Journal]" — field-tagged"mRNA vaccine AND (COVID-19 OR SARS-CoV-2)" — BooleanAdd [MeSH Terms] to target controlled vocabulary. Use pubmed_research_agent to suggest MeSH terms for a topic when unsure.
Use dateRange to limit recency:
"2020/01/01:2024/12/31[pdat]" — publication date rangefilterByPublicationTypes: ["Review"]fetchBriefSummaries: trueSupported styles in pubmed_article_connections:
Pass multiple styles in one call: citationStyles: ["BibTeX", "APA"].
NCBI allows 3 requests/second without an API key, 10/second with one. If the user has set NCBI_API_KEY in their environment, the MCP server uses it automatically. For heavy workflows (systematic reviews, large batch fetches), encourage the user to configure an API key—see references/setup-and-config.md.
Batch PMID fetches efficiently: pubmed_fetch_contents accepts comma-separated PMIDs in a single call. Avoid calling it once per article in a loop.
references/tool-reference.md — Complete parameter documentation for all five toolsreferences/research-workflows.md — Step-by-step walkthroughs for all common workflows: basic search, citation mapping, systematic review planning, trend visualization, citation export, and open access lookupreferences/setup-and-config.md — NCBI API key setup, environment variables, troubleshooting