PubCrawl MCP¶

Primary biomedical sources, directly inside your AI assistant. PubCrawl connects Claude Desktop, Cursor or any MCP client to PubMed, Europe PMC (including preprints), FDA and UK drug labelling, and ClinicalTrials.gov. Every tool is a thin, deterministic wrapper over an official API - nothing invented, every result carrying its PMID, NCT ID or DOI. No API keys required.

Server type: Local stdio (npm)
Auth: None (optional free NCBI key raises PubMed rate limits)
Endpoint: npx -y @pharmatools/pubcrawl
Tools: 14 (literature search, abstracts, full text, drug labelling, US-UK label comparison, trials)
Pricing: Free (MIT)
Category: Analytics & BI
Built by: PharmaTools.AI (github.com/nickjlamb/pubcrawl)

Why This Matters for Operators¶

The tool PubCrawl is known for is compare_labels - the only MCP tool that surfaces US versus UK drug labelling differences. Ask to compare labelling for a drug and it pulls the live US Prescribing Information from openFDA/DailyMed and the UK SmPC from the eMC, then maps equivalent sections side by side. A cardiovascular claim that is on-label in the US can promote an unlicensed indication in the UK; medical copy reviewers live inside that gap, and compare_labels makes it visible instead of assumed.

For everyone else, PubCrawl replaces open browser tabs: literature search with date and article-type filters, structured abstracts, open-access full text, related-article discovery, citation formatting in four styles, and trial lookup by condition, phase and recruitment status.

Tools & Capabilities¶

Tool	Purpose
search_pubmed	Search PubMed with date, article-type and sort filters; returns PMIDs, authors, journals, DOIs
search_europepmc	Search Europe PMC - broader corpus with preprints (bioRxiv, medRxiv) and patents
get_abstract	Structured abstract in labelled sections, with keywords and MeSH terms
get_full_text	Open-access full text from PubMed Central with parsed sections and captions
find_related	Similar articles via PubMed's neighbor algorithm, ranked by relevance
format_citation	APA, Vancouver, Harvard or BibTeX citation formatting
trending_papers	Recent papers on a topic, optionally filtered to high-impact journals
resolve_drug_name	Brand to generic (or reverse) via RxNorm/openFDA, with drug class and indications
get_uspi	US Prescribing Information sections via openFDA, cited to DailyMed
get_smpc	UK Summary of Product Characteristics from the eMC, with numbered sections
compare_labels	Side-by-side US versus UK labelling with equivalent sections paired
search_by_indication	Drugs approved for a condition, with UK availability cross-referenced
search_trials	ClinicalTrials.gov search by condition, intervention, status and phase
get_trial	Full trial record by NCT ID: eligibility, design, arms, outcomes, locations

Installation¶

claude mcp add pubcrawl -- npx -y @pharmatools/pubcrawl

No install step beyond npx; it fetches on first run. An optional free NCBI API key (ncbi.nlm.nih.gov) raises PubMed rate limits for heavy use.

Configuration¶

{
  "mcpServers": {
    "pubcrawl": {
      "command": "npx",
      "args": ["-y", "@pharmatools/pubcrawl"]
    }
  }
}

Business Relevance¶

Medical copy and regulatory reviewers check that claims are on-label in each market with compare_labels instead of manual SmPC reading
Pharma and medtech operators track literature and preprints for a compound with cited, verifiable results
Clinical research teams build trial landscapes with filters for phase, status and intervention
Consultants and analysts produce cited evidence packs in client-ready citation styles

Integration with CorpusIQ¶

PubCrawl covers the scientific record while CorpusIQ covers the business-data layer. A composed workflow: CorpusIQ pulls commercial context through its connectors - revenue, spend, pipeline - while PubCrawl grounds the clinical story in primary sources, and the two feeds join in one assistant session. For market-expansion work, pair CorpusIQ data with PubCrawl's US-UK label comparison to catch on-label versus off-label gaps before copy ships.

Limitations¶

Local stdio server; no hosted endpoint for teams without Node tooling
Biomedical scope only - no general web search or business news
PubMed rate limits apply without a free NCBI key
Young project (15 GitHub stars); literature retrieval is gated by OpenGATE in CI, labelling and trials by an in-repo fidelity benchmark

FAQ¶

Does it need API keys?¶

No. Every tool runs without keys; an optional free NCBI key raises PubMed rate limits. Results always cite their PMID, NCT ID or DOI.

How is compare_labels different from reading the labels myself?¶

It fetches both live labels and maps equivalent sections - US Indications and Usage against UK section 4.1 - so differences are paired side by side, with missing sides explained and truncated sections flagged.

Which clients does it work with?¶

Any MCP client. The quick start documents Claude Desktop; the same npx command works in Cursor, Codex and other MCP-compatible tools.