NCDC Climate Data Online MCP. Pull historical weather records for any location.

You connect your agent directly to NOAA’s massive historical weather and climate archive using this server. It lets you pull actual, measured records—like temperature or rainfall—for any spot in the country over specific stretches of time. Don't worry about the complexity; this setup handles the entire data discovery pipeline for you.

To figure out what data you need, you start with location and scope. If you don’t know the exact coordinates yet, run list_location_categories to see the organizational structure—you can drill down from Country through State or County. Then, use list_locations to grab a list of specific geographic places using their IDs.

For physical monitoring spots, you'll first run list_stations, which shows all active weather stations within an area or category, and then you pull the full metadata for any given site ID with get_station.

Once you’ve nailed down where to look, you figure out what kind of data it is. To see what measurements are even possible—like whether they recorded 'Snowfall' or 'Maximum Temperature'—you run list_data_types, which lists the specific variables available within a larger category. If you want to group those types first, use list_data_categories to browse high-level groups like Precipitation or Temperature.

You can also narrow down how frequently the data was recorded by running list_data_classes, which shows options like Hourly, Daily, or Monthly aggregation.

Next up is defining the source material itself. To get a catalog of the core archives—the actual datasets housing the records—run list_datasets. This gives you names like GHCND or GSOD. If you want to know what’s in one of those specific archives, use get_dataset to fetch detailed information about that NCDC dataset.

So, if your goal is to retrieve the actual measurements—the historical climate records—you pass all these pieces together into the final call: get_climate_data. This tool takes the specific location parameters you found with list_locations, the station ID from get_station, the chosen dataset, and your required date range to pull the measurements.

You can also refine your search by checking what high-level data categories are available via list_data_categories.

Essentially, you use this server's tools in sequence: first, find the location using list_location_categories and list_locations. Second, identify the source archive with list_datasets and get its details via get_dataset. Third, determine the exact metrics you want using list_data_types and define the time scale with list_data_classes. Finally, gather all those pieces—the location, the station metadata, the dataset type, and the specific measurements—and run get_climate_data to get your numbers.