Menü
Search PubMed for biomedical literature via NCBI E-utilities API. Returns titles, authors, DOIs, abstracts.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
| name | pubmed |
| description | Search PubMed for biomedical literature via NCBI E-utilities API. Returns titles, authors, DOIs, abstracts. |
| version | 2.1.0 |
| author | FFFold |
| license | MIT |
| compatibility | claude-code, opencode, openclaw, hermes-agent, aws-codex |
| platforms | ["macos","linux","windows"] |
| metadata | {"hermes":{"tags":["Research","PubMed","Medical","Literature","NCBI","MeSH","Biomedical"],"related_skills":["arxiv","ocr-and-documents"]},"openclaw":{"requires":{"bins":["python3"]}}} |
Search and retrieve biomedical literature from PubMed via NCBI E-utilities API. Auto-converts natural language queries to PubMed syntax using MeSH Terms.
| Action | Command |
|---|---|
| Basic search | python3 scripts/pubmed.py --query '"diabetes"[MeSH]' |
| By date | python3 scripts/pubmed.py --query '"CRISPR"[MeSH]' --sort date |
| More results | python3 scripts/pubmed.py --query '"cancer"' --max-results 10 |
| No cache | python3 scripts/pubmed.py --query '"vaccine"' --no-cache |
scripts/pubmed.py with the converted query.| Parameter | Required | Default | Description |
|---|---|---|---|
--query | Yes | — | PubMed-formatted English query |
--api-key | No | auto | NCBI API key (auto-resolved from env/.env) |
--max-results | No | 5 | Number of results to return |
--sort | No | relevance | relevance or date (newest first) |
--no-cache | No | false | Disable local result caching |
--api-key CLI argumentNCBI_API_KEY → EUTILS_API_KEY → API_KEY.env files: cwd → skill root → ~/.pubmed/.envFree key: https://www.ncbi.nlm.nih.gov/account/settings/ Without key: 3 req/sec. With key: 10 req/sec.
Each result includes:
| Syntax | Meaning | Example |
|---|---|---|
"term"[MeSH] | MeSH heading | "diabetes mellitus"[MeSH] |
"term"[Title/Abstract] | Title or abstract | "machine learning"[Title/Abstract] |
YYYY[DP] | Publication year | 2024[DP] |
"last N years"[Date - Publication] | Relative date | "last 5 years"[Date - Publication] |
Review[pt] | Publication type | Review[pt] |
AND, OR, NOT | Boolean operators | "diabetes"[MeSH] AND "obesity"[MeSH] |
# Recent reviews on a topic
"long covid"[MeSH] AND Review[pt] AND "last 3 years"[Date - Publication]
# Cross-topic intersection
"CRISPR"[MeSH] AND "cancer"[MeSH] AND 2024[DP]
# Broad keyword search (no MeSH)
"artificial intelligence in radiology"
# Specific journal
"Nature"[Journal] AND "gene therapy"[MeSH]
NCBI E-utilities occasionally returns SSL: UNEXPECTED_EOF_WHILE_READING for complex MeSH queries with multiple Boolean operators. Simple keyword queries are more reliable.
Workaround: If a complex MeSH query fails, simplify it or retry. The script handles this gracefully and returns an error message.
| API | Without Key | With Key | Auth |
|---|---|---|---|
| esearch/esummary | 3 req/s (5 burst) | 10 req/s | Optional |
| efetch | 3 req/s (5 burst) | 10 req/s | Optional |
HTTP 429 is returned on limit hit — the script does not auto-retry.
The script uses regex-based XML parsing for abstracts (no external deps). Edge cases:
sort=date returns the most recently indexed papers, not necessarily the most recently published. A paper published in 2024 but indexed in 2026 may appear first.
/tmp/pubmed_cache/--no-cache to bypassAfter running a search, verify:
Use scripts/pubmed.py for the full workflow. No external dependencies — Python stdlib only.
# Install (copy to a known location)
cp scripts/pubmed.py /usr/local/bin/pubmed-search
# Or run directly
python3 scripts/pubmed.py --query '"diabetes"[MeSH]' --sort date --max-results 5