PubMed Database Search

Problem

Biomedical literature searches often get stuck in query syntax, MeSH subheadings, publication-type filters, and API retrieval. Manually composing AND, OR, NOT, field tags, and date ranges can miss synonyms or misuse automatic term mapping. Scripted retrieval of abstracts, PMIDs, and DOIs also requires handling rate limits and history servers.

How the skill works

The skill structures PubMed retrieval into reusable engineering steps: build queries with Boolean operators, field tags, phrases, and wildcards, such as [tiab], [mh], [pt], and [dp]; increase precision with MeSH terms and subheadings like diabetes mellitus/therapy[mh]; then filter by article type, date, and free full text[sb]. For automation, it covers NCBI E-utilities endpoints including ESearch, EFetch, ESummary, EPost, and ELink, and recommends usehistory=y, local caching, and exponential backoff for large result sets. It also supports systematic review searches, citation matching, similar-article discovery, and exporting to .nbib, CSV, or XML.

Boundaries

It fits biomedical and life-science literature search, but it does not provide paid publisher full text. Displayed results are capped at 10,000 items, search history and clipboards expire, and APIs remain subject to 3 to 10 requests per second. For complex workflows, consult its api_reference.md, search_syntax.md, and common_queries.md as needed.