The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Nih Supplements listing page.
NIH fact sheets for vitamins, minerals and a few botanicals — uses, safety, interactions with medications, references — plus real supplement product labels from NIH's Dietary Supplement Label Database, and NCCIH's herb safety/science summaries for the many popular herbs ODS itself doesn't cover.
Part of Pipeworx — an MCP gateway connecting AI agents to 1683+ live data sources.
supplement_factsheet(name, level?) — a topic's ODS fact sheet: what it
does, whether it works, safety, and interactions with medications.
level is consumer (short, default) or professional (long, clinical,
with a references list).supplement_factsheets(query?) — the bundled list of ~45 topics ODS
publishes a fact sheet for, and which reading levels each has.supplement_labels(query, limit?) — search real product labels (brand,
ingredients, amounts, claims) from the Dietary Supplement Label Database
(DSLD) by product, brand, or ingredient.supplement_label(id) — one product's full label detail: ingredient
amounts, other ingredients, manufacturer contact, precaution statements.herb_at_a_glance(herb) — NCCIH's "Herbs at a Glance" science-and-safety
summary for ~56 popular herbs (turmeric, ginkgo, echinacea, garlic,
ginseng, milk thistle, St. John's Wort, and more) that ODS does not have
its own fact sheet for.supplement_recent_changes(since, limit?) — new DSLD product labels
entered on or after a date, newest first, deterministic on replay. DSLD is
the only NIH supplement upstream with a queryable change signal; the
response states plainly that withdrawals (offMarket transitions) and
ODS/NCCIH content updates are NOT diffable upstream rather than pretending
to cover them.supplement_recent_changes(since) pages DSLD newest-first
(sort_by=entryDate&sort_order=desc) to the cutoff — deterministic, so
replays are idempotent until NIH enters new labels.lastmod (checked 2026-09-08; NCCIH's sitemap
exists but carries no dates, and HHS's syndication API is bot-walled).
Each supplement_factsheet response carries its own reviewed date.supplement_factsheets() lists all ~45 ODS topics; the ~56 NCCIH herbs
are in NCCIH_HERBS in src/index.ts (and any name from either list can
be fetched in full). DSLD's full corpus (~215k labels) is enumerable via
the same sorted paging (size/from), and NIH also publishes full DSLD
dataset downloads at https://dsld.od.nih.gov/ for bulk use. There is no
versioned bulk file for the fact sheets — enumeration over the bundled
inventory IS the snapshot path for them.Keyless. All three upstreams (ODS fact sheets, DSLD, NCCIH) are open NIH sites with no API key.
?resourcename=<slug>&readinglevel=Consumer|HealthProfessional&outputformat=XML).
Trap: a fresh call from a browser-shaped User-Agent returns real XML; a
rapid second call from the same client comes back HTTP 200 with an HTML
bot-check page instead of data — same status code, no error, completely
different body. Never trust an HTML-shaped 200 from this host as a "not
found"; it means "slow down and retry". This pack handles it with a
per-isolate cache plus a floor between outbound calls
(spacedFetch/MIN_GAP_MS in src/index.ts) — don't remove that if you
touch this file.ODS_TOPICS in
src/index.ts) is bundled by hand from this page, read once on 2026-09-07.
Almost every topic uses the same slug for both reading levels — level is a
separate query param — except two: "Dietary Supplements in the Time of
COVID-19" and Chromium each use a different slug per level, which is why
the inventory stores consumerSlug/professionalSlug per topic instead of
one shared slug.hit._source carries brandName, fullName, productType,
allIngredients, claims, netContents, offMarket._id field on 2026-09-07). Carries ingredient amounts
(ingredientRows[].quantity[]), other ingredients, manufacturer contact,
and the precaution/warning statements printed on the bottle.__NEXT_DATA__, or XHR/API was found (checked
2026-09-07), but every page's content <h2> carries a stable id
attribute — background, how-much-do-we-know, what-have-we-learned,
what-do-we-know-about-safety, keep-in-mind, for-more-information,
key-references — verified identical and in the same order across 5 herbs
(ashwagandha, echinacea, garlic, ginkgo, turmeric). That is the "sections
are stable" condition required before building a bounded prose extractor
over these pages (docs/herbal-sources-scope.md §2); it passed, so
herb_at_a_glance ships. The ~56-herb list (NCCIH_HERBS) is likewise
bundled by hand from nccih.nih.gov/health/herbsataglance, read once on
2026-09-07 — there is no live index for this either.Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):
tools/list at https://gateway.pipeworx.io/nih-supplements/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:
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 account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/supplement_factsheet. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.
This package also runs as a local stdio MCP server — no Pipeworx account, no gateway round-trip:
Or run it directly to confirm it starts:
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.
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:
The gateway picks the right tool and fills the arguments automatically.
MIT