NOAA Climate MCP. Analyze climate trends from raw data to 30-year normals.

The NOAA Climate — Historical Weather Records MCP Server gives your agent direct access to the largest archive of daily weather data on the planet, pulling raw records instead of just vague reports. You can pull temperature and precipitation readings for over 100,000 stations worldwide. Before you do anything else, you need a station ID; that's where search_stations comes in play.

You use this tool to identify the unique NOAA station ID, name, and coordinates by searching using a location bounding box or just keywords.

Once you have that crucial station identifier, you can define what 'normal' looks like for any spot using get_climate_normals. This fetches the 30-year statistical baseline—the established standard used to measure if current conditions are deviating from average. It sets the benchmark against which you compare everything else.

For granular analysis, you'll use get_daily_data. This tool pulls precise daily records for a specified station and date range. You get max/min temperatures, total precipitation measurements, snow depth readings, and wind data—all highly detailed, day-by-day metrics that let you see the true variability of the weather. If you're doing deep trend analysis, or comparing specific historical days to today's reading, this is the dataset you want.

If raw daily noise overwhelms your goal, you can aggregate the data. get_monthly_summary generates monthly totals and averages. This gives you an average temperature, total precipitation, and heating degree days for a given month, perfect when you're analyzing seasonal patterns without getting bogged down in day-to-day fluctuations.

For long-term climate reporting that spans multiple decades, get_yearly_summary provides the necessary scope. This tool gives yearly averages and extreme values, letting you track massive shifts in climate data over time. You’ll use this when your report needs to cover a decade or more of Earth's weather history.

In short, you first gotta locate the station ID using search_stations. Then, if you need context, you check the baseline with get_climate_normals. If you want deep detail, you run get_daily_data for specific dates. To spot seasonal trends, use get_monthly_summary. And finally, to build those multi-decade reports, grab yearly metrics using get_yearly_summary.

You're pulling hard data points; the system doesn't just give you a pretty summary.