cmer81/open-meteo-mcp/open-meteo

Open-Meteo MCP

Overview

This MCP server provides direct access to Open-Meteo weather APIs via dedicated tools. No direct API calls needed — call the appropriate tool with coordinates and the variables you want.

If you only have a city name, call geocoding first to obtain coordinates and timezone.

Which Tool to Use

Key Parameters

geocoding

Returns: latitude, longitude, timezone, country. Attribution: geocoding data from GeoNames.

weather_forecast

\At least one of `hourly`, `daily`, or `current` is required. \\*Always set timezone=auto when requesting daily variables.

One model per request. For multi-model comparison, use open-meteo-advanced model tools in parallel.

Common hourly variables: temperature_2m, relative_humidity_2m, apparent_temperature, precipitation, precipitation_probability, wind_speed_10m, wind_direction_10m, weather_code, cloud_cover, uv_index, visibility, is_day. Upper-air pressure-level variables are also available, e.g. temperature_850hPa, wind_speed_500hPa, geopotential_height_200hPa (levels: 1000 down to 30 hPa).

Common daily variables: temperature_2m_max, temperature_2m_min, apparent_temperature_max, precipitation_sum, precipitation_probability_max, wind_speed_10m_max, wind_gusts_10m_max, weather_code, sunrise, sunset, uv_index_max, shortwave_radiation_sum

weather_archive

Same parameters as weather_forecast, with these differences:

Use past_days on weather_forecast for recent history (up to 92 days back). Use weather_archive for older dates.

air_quality

\*At least one of hourly or current is required.

Common variables: pm2_5, pm10, european_aqi, us_aqi, carbon_monoxide, nitrogen_dioxide, ozone, sulphur_dioxide, dust, uv_index, alder_pollen, birch_pollen, grass_pollen, mugwort_pollen, olive_pollen, ragweed_pollen

Attribution: air quality data from CAMS (Copernicus Atmosphere Monitoring Service).

marine_weather

\*At least one of hourly, current, or daily is required.

Common hourly variables: wave_height, wave_direction, wave_period, wind_wave_height, wind_wave_direction, wind_wave_period, swell_wave_height, swell_wave_direction, swell_wave_period, sea_surface_temperature, ocean_current_velocity, ocean_current_direction

Common daily variables: wave_height_max, wave_direction_dominant, wave_period_max, swell_wave_height_max

flood_forecast

Variables: river_discharge (deterministic), river_discharge_mean, river_discharge_median, river_discharge_max, river_discharge_min, river_discharge_p25, river_discharge_p75

Data source: GloFAS (Global Flood Awareness System).

elevation

Returns altitude in metres (one value per coordinate pair).

Examples

"What is the weather like in Lyon tomorrow?"

1. geocoding with name: "Lyon" → latitude, longitude, timezone
2. weather_forecast with those coordinates + daily: ["temperature_2m_max", "temperature_2m_min", "precipitation_sum", "weather_code"], forecast_days: 2, timezone: "auto"

"Is the air quality good in Paris right now?"

1. geocoding with name: "Paris" → coordinates
2. air_quality with coordinates + current: ["european_aqi", "pm2_5", "pm10"]

"What was the average temperature in Berlin in July 2024?"

1. geocoding with name: "Berlin" → coordinates, timezone
2. weather_archive with coordinates + daily: ["temperature_2m_max", "temperature_2m_min"], start_date: "2024-07-01", end_date: "2024-07-31", timezone: "auto"

"What are the wave conditions near Biarritz this weekend?"

1. geocoding with name: "Biarritz" → coordinates
2. marine_weather with coordinates + hourly: ["wave_height", "wave_direction", "wave_period", "swell_wave_height"], forecast_days: 3

Best Practices

• Always geocode first if you only have a city or place name.
• Always set `timezone=auto` when requesting daily variables — otherwise results are in UTC and day boundaries will be wrong for non-UTC locations.
• Request only the variables you need — smaller payloads are faster and easier to parse.
• Use `current` for instant conditions, hourly for time series, daily for day-level summaries.
• `forecast_days: 1` returns today only; forecast_days: 2 returns today + tomorrow.
• For multi-model comparison or forecasts beyond 16 days, use open-meteo-advanced.