Summary
The explicit parameter list is the design decision that matters: nothing comes back that you did not name, which keeps a forecast small enough to reason about inside a session.
All three tools repeat that instruction, and they differ mainly in which family of parameters they accept and which model they can run against. get_weather_forecast carries the extra shape — levels for height and temp_unit for the scale — while the wave and air quality calls stay at lat, lon, model and parameters. There is no place lookup anywhere here, so the caller has to arrive with a coordinate already in hand.
What it is
A client for Windy's point forecast API, split by forecast family into weather, ocean wave, and air quality with pollen.
What you get
- get_weather_forecast returns a forecast time series at a lat and lon from the model you name, with levels selecting heights and temp_unit fixing the scale.
- get_wave_forecast returns an ocean wave time series for a coordinate from a chosen model.
- get_air_quality_forecast returns air quality and pollen time series for a coordinate.
- Only the parameters you explicitly list are fetched, so a response carries the variables you asked for and nothing beside them.
Requirements
A Windy API key, supplied as WINDY_API_KEY, and the coordinate you want the forecast for — every call takes lat and lon.
Setup effort
One command plus a key — uvx windy-mcp-server, then supply credentials