# Weather MCP Server by Meteosource > Forecasts, severe-weather alerts, air quality, 20-year weather history, solar and wind production. - name: com.meteosource/weather - version: 0.1.2 - connection class: R1 (autonomous) - trust: 100/100 - quality: 80/100 (good) - owner: not claimed - descriptor: https://api.protogrid.dev/v1/servers/com.meteosource%2Fweather - tools: https://api.protogrid.dev/v1/servers/com.meteosource%2Fweather/tools **An agent can connect on its own if it already holds `api_key`.** Remote endpoint behind a static API key or header. The registry never sees the value; the agent fills the placeholder. ## Connection (mcpServers) ```json { "mcpServers": { "weather": { "type": "http", "url": "https://mcp.meteosource.com/mcp", "headers": { "Authorization": "Bearer ${api_key}" } } } } ``` Placeholders to fill: `${api_key}`. ## Trust - hygiene: 100 - liveness: 100 - freshness: 100 - provenance: 100 - drivers: +repository +namespace-matches-repo +dns-namespace +website +reachable +uptime-100% +updated-12d-ago - Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed. ## Quality 80/100 (good) ### protocol: 88 (weight 25) - pass `protocol.modern`: supports 2026-07-28 (server/discover) - pass `protocol.stateless`: answers without a session - pass `protocol.transport`: streamable HTTP - warn `protocol.list_ttl`: tools/list declares ttlMs 0, so clients re-fetch it on every use ### authorization: n/a (weight 15) - n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth - n/a `auth.secret_in_url`: no templated URL ### tool hygiene: 64 (weight 30) - pass `tools.descriptions`: every tool has a description - warn `tools.description_length`: 4 tools have descriptions over 1,024 characters, which crowds the context - pass `tools.schemas`: every tool has an object input schema - fail `tools.annotations`: no tool declares readOnlyHint or destructiveHint, so clients cannot tell safe calls from risky ones - fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: get_forecast (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); get_alerts (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); get_wind_production (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); … - pass `tools.token_cost`: about 4,606 tokens to load every tool - pass `tools.api_dump`: tools are not a one-to-one API dump ### stability: 100 (weight 15) - pass `stability.changes`: no tool changes in 30 days - pass `stability.rug_pull`: no tool changed its meaning under the same name ### dependencies: n/a (weight 15) - n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package - tools: 7, about 4,606 tokens to load them all - tool set hash: acde57f4539435dc54f7cace65a0d840 - badge: [![protogrid quality](https://protogrid.dev/badge/com.meteosource/weather.svg)](https://protogrid.dev/servers/com.meteosource/weather) - Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed. ### Tool changes _No tool definition changes recorded._ ## Remotes - https://mcp.meteosource.com/mcp (streamable-http, auth api_key ${api_key}, reachable true, uptime30d 1) ## Packages _none_ ## Tools (7) - `get_air_quality`: Hourly air-quality (smog) forecast for up to 20 locations, about 4 days ahead. Columns: `air_quality` (index, 1 = good, higher = worse) and pollutant surface concentrations — default pm10, pm25, no2_surface, so2_surface, ozone_surface; also available via `variables`: co_surface, no_surface, ozone_total, aerosol_550, dust_550nm, dust_mixing_ratio_05. Output is a table per location (`columns` once, `rows` as arrays), trimmed to `hours`. Not available on the free tier. - `get_alerts`: Official weather warnings (severe weather alerts) for up to 20 locations. Returns only locations that currently have an active alert, with event type, severity, validity window and headline — an empty list means all clear. Cheap to call frequently; designed for monitoring/polling. Coverage: USA, EU countries and Canada. Requires the startup tier or higher (not available on the free tier). - `get_forecast`: Weather forecast for up to 20 locations in one call. Sections (comma-separated): 'current' (now), 'hourly' (up to 168 h ahead, trimmed to `hours`), 'daily' (up to 30 days ahead, trimmed to `days`), 'minutely' (precipitation minute-by-minute, higher tiers), 'alerts' (official weather warnings, startup tier and above), or 'all'. Use `variables` to keep only specific fields (e.g. ["temperature", "wind", "precipitation", "probability", "irradiance"]) and reduce output size — recommended when querying many locations. Timezone defaults to each location's local time; units default per-country ('metric' forces SI). Forecast horizon depends on the subscription tier (free: 7 days). Output: time series (hourly, daily, minutely) are tables — `columns` listed once and `rows` as arrays in column order; nested API groups are flattened with underscores (wind_speed, probability_storm, cloud_cover_total). `units` is given once at the top level when it is the same for every location. - `get_solar_production`: Production of photovoltaic systems, past and forecast, for up to 20 sites. Each site is lat + lon plus the installation: module_kw (installed kWp; use 1 to get specific yield per kWp), inverter_kw (defaults to module_kw), tilt (degrees from horizontal, default 30) and orientation (azimuth the modules face, clockwise from north: 180 = south, the default). Dates are YYYY-MM-DD, inclusive, from up to 20 years back to 4 days ahead, so one call can cover last week's actual weather and the coming days. The renewables plan includes the last 3 years; older days need the paid historical add-on, otherwise the API refuses them ("historical data this old"), reported per site. Granularity: 'hourly' (power at each full hour, max 14 days per call), 'daily' or 'monthly' (energy, max 366 days per call — call repeatedly for multi-year yield; past days are cached so repeats are cheap). Columns: production_kw = output of the system at that hour (an instantaneous value); production_kwh = energy of the day or month, summed from those hourly values; production_max_kw = its peak. `variables` adds the weather behind the production: irradiance (ghi, dni, dhi in W/m2 hourly; ghi/dni/dhi_sum_wh_m2 daily and monthly), temperature, cloud_cover, humidity, dew_point, pressure, snowfall (flag per hour; snowfall_hours daily and monthly). Each site+day costs one API request on first fetch; a single call is capped at 400 site-days — split longer ranges into multiple calls. Needs a key on the renewables tier. Output: `columns` once at the top level; each site has `rows` as arrays in column order. - `get_weather_history`: Historical weather (archive) with server-side aggregation. The archive holds 20 years, but how much of it a key may read depends on its plan: startup includes the last 7 days, standard the last year, premium and flexi all 20 years; byvalue reaches 366 days back and no further. Older days need Meteosource's paid historical add-on (billed per location and day); without it the API refuses such a day with "not allowed to access historical data this old", reported per location, so start inside the plan's range. Dates are YYYY-MM-DD, inclusive. Granularity: 'hourly' (raw hours, max 14 days per call), 'daily' (per-day min/max/avg/sum statistics, max 366 days) or 'monthly' (folded from daily, max 366 days per call — call repeatedly for multi-year analyses; results are cached so repeats are cheap). Variables (default: temperature, precipitation, wind): cape, cloud_cover, dew_point, evaporation, feels_like, humidity, irradiance, ozone, precipitation, precipitation_type, pressure, soil_temperature, surface_temperature, temperature, weather, wind, wind_distribution. Aggregations per variable: temperature min/max/avg, precipitation sum, wind speed avg + gusts max, irradiance sum in Wh/m2 (useful for PV analysis), weather = most frequent code of the day, precipitation_types = kinds seen that day (e.g. "rain,snow"; "none" when dry) — use weather and precipitation_type to document what happened (snow, fog, thunderstorm). wind_distribution (daily/monthly; for wind-turbine siting) = hours per 10 m wind-speed bin (wind_hours_0_2 … wind_hours_10_plus) and wind_power_density_w_m2 = 1/2 * 1.225 * mean(v^3) at 10 m — scale to hub height yourself (the API has no hub-height wind). `normals=true` (daily/monthly) adds long-term columns for each calendar day: temperature_normal_avg/min/max, temperature_record_min/max, wind_normal_speed, wind_record_gusts, precipitation_normal_sum and precipitation_normal_probability — compare a period with its normal in one call. Each location+day costs one API request on first fetch; a single call is capped at 400 location-days — split longer ranges into multiple calls. Not available on the free tier (startup+). Output: `columns` once at the top level; each location has `rows` as arrays in column order (a year of daily rows is ~15 kB). - `get_wind_production`: Production of wind farms, past and forecast, for up to 20 sites. Each site is lat + lon plus the farm: turbine_model (catalogue below), exactly one of turbine_count or total_capacity (MW), hub_height (metres, default 100) and roughness of the surrounding terrain (water, open, farmland — the default, suburban, urban). Dates are YYYY-MM-DD, inclusive, from up to 20 years back to 4 days ahead, so one call can cover last week's actual weather and the coming days. The renewables plan includes the last 3 years; older days need the paid historical add-on, otherwise the API refuses them ("historical data this old"), reported per site. Granularity: 'hourly' (power at each full hour, max 14 days per call), 'daily' or 'monthly' (energy, max 366 days per call — call repeatedly for multi-year yield; past days are cached so repeats are cheap). Columns: production_mw = output of the whole farm at that hour (an instantaneous value); production_mwh = energy of the day or month, summed from those hourly values; production_max_mw = its peak. `variables` adds the weather behind the production: wind (10 m above ground, not hub height: wind_10m_speed, wind_10m_gusts and wind_10m_angle hourly; speed avg/max and gusts max daily and monthly), temperature, dew_point, pressure. Turbine catalogue (series + rotor diameter in m / rated power in kW): Adwen AD116/5000; Enercon E48/800, E-53/800, E-70/2000, E-70/2300, E-82/2000, E-82/2300, E-82/2350, E-82/3000, E-92/2350, E-101/3050, E-101/3500, E-115/3000, E-115/3200, E-126/4200, E-126/7500, E-126/7580, E-141/4200; Eno ENO100/2200, ENO114/3500, ENO126/3500; GE GE100/2500, GE103/2750, GE120/2500, GE120/2750, GE130/3200; Nordex N90/2500, N100/2500, N117/2400, N131/3000, N131/3300, N131/3600; Senvion MM92/2050, MM100/2000, S104/3400, S114/3200, S114/3400, S122/3000, S122/3200, S126/6150, S152/6330; aerodyn SCD168/8000; Siemens SWT113/2300, SWT113/3200, SWT120/3600 (the default), SWT130/3300, SWT130/3600, SWT142/3150; Vestas V80/2000, V90/2000, V90/2000/GS, V90/3000, V100/1800, V100/1800/GS, V112/3000, V112/3075, V112/3300, V112/3450, V117/3300, V117/3450, V117/3600, V126/3000, V126/3300, V126/3450, V164/8000, V164/9500; Vensys VS112/2500. Each site+day costs one API request on first fetch; a single call is capped at 400 site-days — split longer ranges into multiple calls. Needs a key on the renewables tier. Output: `columns` once at the top level; each site has `rows` as arrays in column order. - `search_places`: Find locations and their Meteosource place_id. Use this first to resolve city names, postal codes or coordinates into place_id values for the other weather tools. Provide either `query` (place name or postal code) to search, or `lat` + `lon` to find the nearest known place. Returns up to 10 matches ordered by relevance, each with place_id, administrative area, country, coordinates and timezone. Note: lat/lon are returned in degree-letter format (e.g. "50.08804N").