NOAA Public Data MCP Server

What does NOAA Public Data MCP Server do?

Four NOAA search sites, zero manual browsing: NOAA Public Data MCP Server turns data.noaa.gov/onestop, ncei.noaa.gov/search/content, ncei.noaa.gov/access/search/dataset-search, and ncei.noaa.gov/stormevents into MCP tools any AI agent can call directly. Ask for "Gulf of Mexico sea surface temperature datasets from 2020" or "Texas tornadoes in January 2020" in plain language and get back real, structured results with working links — not a screen-scrape of a form you have to fill in yourself.

Each of the four tools covers that source's real search/filter surface — full-text query, facets, keywords, spatial bounding box, date ranges, event type/severity, sort, pagination — reverse-engineered from the actual site (documented JSON API where NOAA has one, clean HTML parsing where it doesn't). Results carry links to dataset landing pages, access servers, and event detail pages. The Actor never downloads a dataset file or PDF — it only tells you where they are, so you stay lightweight and in control of what you actually fetch.

Because it runs on Apify in Standby mode, it behaves like a normal always-on HTTP MCP endpoint — no cold start per query — plus you get Apify's monitoring, scheduling, and proxy infrastructure for free if you need it.

Why use NOAA Public Data MCP Server?

• Skip four different websites and four different query languages. One consistent tool interface instead of learning OneStop's facet JSON, NCEI's REST params, a Drupal search form, and a JSP CGI form.
• Agent-ready by design. Point Claude (or any MCP client) at real NOAA data during research, reporting, or analysis — no copy-pasting URLs, no manual form-filling, no HTML to eyeball.
• Structured, linked, lightweight. Every result is JSON with the actual landing-page/access/detail links attached — pull exactly the dataset or event you need, on your own terms, without the Actor pre-downloading anything for you.
• Built for real workflows. Pull Storm Events Database records (tornadoes, floods, hail, etc.) for a state/date range straight into an agent workflow, dashboard, or notebook.

How to use NOAA Public Data MCP Server

1. Deploy this Actor to Apify (or use the published one) with Standby mode enabled.

2. Add it to your MCP client as an HTTP/Streamable-HTTP server:

   claude mcp add --transport http noaa-data \
     https://rl1987--noaa-data-mcp.apify.actor/mcp \
     --header "Authorization: Bearer $APIFY_TOKEN"
   

   The token is your own Apify API token (Console → Settings → Integrations), not the Actor owner's — export it as $APIFY_TOKEN in your shell, or substitute your own.

3. Verify the connection and tool count:

   claude mcp list
   # or, inside a Claude Code session:
   /mcp
   

4. Ask your agent to search, e.g. "Use noaa-data to find OneStop datasets about hurricanes with a THREDDS download link" or "List Texas tornado events from January 2020 via search_storm_events."

A team can commit a .mcp.json with just the URL (no token) to a shared repo; each teammate runs claude mcp add --scope local once with their own token.

Tools

Tool	Source	Covers
search_onestop_data	OneStop	full-text query, facet filters (science, locations, instruments, platforms, projects, data centers, data/file formats, link protocols), date range, bounding box, collection/granule search, pagination
search_ncei_dataset_search	NCEI Dataset Search	full-text query, date range, bounding box, GCMD keywords, pagination
search_ncei_content	NCEI Content Search	full-text query, content type, topic filters, sort, pagination
search_storm_events	Storm Events Database	state, event type, date range, hail/tornado/wind severity filters, sort

Every tool returns JSON with result metadata and a links/detailUrl field pointing at the real NOAA page or access server — datasets and PDFs are never fetched or stored by this Actor.

Example output (search_storm_events, abbreviated)

{
  "source": "https://www.ncei.noaa.gov/stormevents/",
  "count": 14,
  "truncated": false,
  "results": [
    {
      "EVENT_ID": "873054",
      "CZ_NAME_STR": "WISE CO.",
      "BEGIN_DATE": "01/10/2020",
      "EVENT_TYPE": "Tornado",
      "TOR_F_SCALE": "EF0",
      "STATE_ABBR": "TX",
      "detailUrl": "https://www.ncei.noaa.gov/stormevents/eventdetails.jsp?id=873054"
    }
  ]
}

Pricing / cost

This Actor uses Apify's Pay Per Event model — see .actor/pay_per_event.json for the configured tool-call price. You're charged once per completed tool call, not per underlying HTTP request the tool makes internally.

Tips

• search_storm_events caps at 500 records (NOAA's own limit) — narrow by state, event type, or a shorter date range if you hit it.
• For OneStop, valid facet values (e.g. under science, dataFormats) are visible in the facets field of a prior, broader query.
• Use search_type='granule' with collection_id on search_onestop_data to list files within a specific dataset collection.

FAQ, disclaimers and support

All data queried here is public NOAA/NCEI information. This Actor only relays search results and links; it does not mirror, cache, or redistribute NOAA datasets. Respect NOAA's own terms of use for anything you download via the returned links. Found a bug or need a custom variant? Open an issue on the Actor's Issues tab.