Florida GIO Open Data MCP Server

Florida GIO Open Data MCP Server

An MCP server that lets AI agents search and query the Florida GIO open data portal — the Florida Geographic Information Office's public catalog of ~200 state geospatial datasets (boundaries, environment, transportation, imagery, hydrography, and more).

It talks directly to the portal's public DCAT-US catalog feed for dataset discovery and to each dataset's own Esri ArcGIS GeoServices REST API for live feature data, so setup is zero-config: no Florida GIO account, no API key.

Coverage

Tool	What it does
search_datasets	Keyword search across the full catalog (title, description, tags)
get_dataset	Full metadata for one dataset, incl. every available download format (GeoJSON, Shapefile, CSV, KML, File Geodatabase, GeoPackage, Excel, ...)
get_layer_info	Field names, types, and geometry type for an ArcGIS FeatureServer/MapServer layer
query_features	Starts a feature query against a layer (SQL-92 where clause and/or spatial filter); returns a job_id
get_query_result	Polls a query_features job; returns the result as GeoJSON once done

Typical flow: search_datasets("wetlands") → get_dataset(id) to get the featureServiceUrl → get_layer_info(url) to see field names → query_features(url, where="county='LEON'") to start the query → get_query_result(job_id) to fetch matching records.

Caveats

• The catalog is cached in-memory for 15 minutes per Actor instance to avoid re-fetching the ~200-dataset feed on every search_datasets/get_dataset call.
• query_features runs asynchronously because some ArcGIS layers (especially large statewide ones) can take well over a minute to answer a filtered query — synchronous MCP tool calls would otherwise block or time out. Poll get_query_result until status is done or error. Job results are kept for 30 minutes.
• query_features/get_layer_info proxy straight to the dataset's own ArcGIS server — result caps, field names, and query support (e.g. spatial filters) vary by dataset the same way they would if you queried the ArcGIS REST endpoint directly. Large unindexed text-field filters (e.g. a city-name where clause on a statewide parcel layer) can be slow or fail server-side; a spatial (geometry) filter combined with an indexed numeric field is usually faster.
• Some catalog entries are external federal layers (e.g. NOAA-hosted) surfaced through the Florida GIO portal rather than Florida-hosted data — get_dataset shows the true featureServiceUrl host either way.

How to connect

Find the Actor's Standby endpoint URL on its API tab in Apify Console (shape: https://rl1987--fl-gio-mcp.apify.actor/mcp).

Add it with the Claude Code CLI:

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

$APIFY_TOKEN is your own Apify API token, not the Actor owner's. Verify the connection and tool count with claude mcp list or /mcp inside Claude Code.

A team can share a .mcp.json with just the URL committed to a repo; each teammate runs claude mcp add --scope local once with their own token.

Output shape

search_datasets returns:

{
  "total": 119,
  "results": [
    {
      "id": "https://www.arcgis.com/home/item.html?id=...",
      "title": "Water Level Prediction Stations",
      "landingPage": "https://geodata.floridagio.gov/datasets/fedmaps::water-level-prediction-stations",
      "keyword": ["water level", "tide", "..."],
      "modified": "2026-08-10T05:52:41.962Z",
      "spatial": "-179.81,-29.25,179.97,71.36",
      "featureServiceUrl": "https://services2.arcgis.com/.../FeatureServer/0"
    }
  ]
}

get_dataset returns the raw DCAT-US record for that item (all distribution download links included). get_layer_info returns {name, geometryType, fields}. query_features returns {job_id, status: "running"}. get_query_result returns {status: "running"} while pending, or once done {status: "done", featureCollection, truncated, feature_count} where featureCollection is a standard GeoJSON FeatureCollection.

Pricing

Pay-per-event: see .actor/pay_per_event.json for the current price per completed tool call (tool-call event) — for query_features, the charge fires once the job completes successfully, not when it's started. Failed calls are not charged.

Data source

All data is sourced live from Florida GIO: the DCAT-US catalog feed for discovery, and each dataset's own ArcGIS GeoServices REST API for feature queries. Refer to the portal for licensing terms of individual datasets.

Resources

• What is Anthropic's Model Context Protocol?
• How to use MCP with Apify Actors
• FastMCP documentation
• Apify SDK documentation
• ArcGIS REST API reference

Did you find this useful?

⭐ Rate this actor on Apify! Your feedback helps other users find it and helps us keep improving it.