Ncbi Datasets MCP Server

@pipeworx/ncbi-datasets

NCBI Datasets (US National Library of Medicine) — which genome assemblies exist for an organism and how good they are, where a gene sits and what it is called in every other database, and the NCBI taxonomy tree with per-node assembly and gene counts.

Part of Pipeworx — an MCP gateway connecting AI agents to 1683+ live data sources.

Tools

• ncbi_genome_reports(taxon?, accession?, reference_only?, assembly_level?, limit?) — assemblies with level, length, contig/scaffold N50, GC%, submitter, BioProject and the paired RefSeq/GenBank accession.
• ncbi_gene_by_symbol(symbols, taxon?) — Entrez gene id, biotype, chromosome, RefSeqGene coordinates, plus HGNC / Ensembl / UniProt / OMIM cross-references and every synonym.
• ncbi_taxonomy(taxons) — resolve a tax id, scientific name or common name to the full lineage, children, rank and live assembly/gene counts.

Auth

Keyless. NCBI rate-limits anonymous callers by IP at roughly 5 requests/second across api.ncbi.nlm.nih.gov; the pack sends an identifying User-Agent, does not retry, and surfaces a 429 with that explanation rather than looping.

Not the same service as ncbi-eutils

E-utilities is a generic search/fetch layer over ~40 Entrez databases that hands back records you must parse. Datasets answers structured questions about genomes, genes and taxonomy with structured rows. Use this pack for those three; use ncbi-eutils for PubMed, dbSNP and the rest of Entrez.

Data sources

• https://api.ncbi.nlm.nih.gov/datasets/v2/genome/taxon/{taxon}/dataset_report
• https://api.ncbi.nlm.nih.gov/datasets/v2/genome/accession/{acc}/dataset_report
• https://api.ncbi.nlm.nih.gov/datasets/v2/gene/symbol/{symbols}/taxon/{taxon}
• https://api.ncbi.nlm.nih.gov/datasets/v2/taxonomy/taxon/{taxons}

Things that cost time to rediscover (measured 2026-09-17)

• Without filters, a genome report is every assembly ever deposited. Taxon 562 (E. coli) has 492,559, and page one of two is a GenBank/RefSeq pair of the same assembly — technically correct, practically useless. filters.reference_only=true takes that to 2, so ncbi_genome_reports defaults to it for a taxon lookup and reports that it did.
• Every assembly exists twice, as GCA_ (GenBank) and GCF_ (RefSeq), cross-linked by paired_accession. Counting rows double-counts assemblies.
• Multi-value paths do not preserve order. taxonomy/taxon/9606,10090 returns Mus musculus first. Match on the returned query / symbol, never on position.
• Numeric fields are strings in gene reports ("gene_id":"672") but numbers in taxonomy reports (tax_id: 9606). Both are passed through as received.
• A taxon path accepts a tax id, a scientific name or a common name — 9606, Homo sapiens and human all work.

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "ncbi-datasets": {
      "url": "https://gateway.pipeworx.io/ncbi-datasets/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/ncbi-datasets/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1683+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

No MCP client? Call it over HTTP

curl -X POST https://gateway.pipeworx.io/v1/tools/ncbi_genome_reports \
  -H 'Content-Type: application/json' \
  -d '{"taxon":"9606","limit":3}'

No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/ncbi_genome_reports. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.

Standalone (no gateway account)

This package also runs as a local stdio MCP server — no Pipeworx account, no gateway round-trip:

{
  "mcpServers": {
    "ncbi-datasets": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-ncbi-datasets"]
    }
  }
}

Or run it directly to confirm it starts:

npx -y @pipeworx/mcp-ncbi-datasets

It speaks MCP over stdin/stdout and answers initialize/tools/list/tools/call for only this pack's tools — none of the shared meta-tools the gateway connection above adds. Same source, same tools, no ask_pipeworx routing.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Ncbi Datasets data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides
• pipeworx.io

License

MIT