AI · Updated for 2026

Meteoprog API MCP Server: Weather Data for AI Assistants

Meteoprog API MCP Server lets ChatGPT, Codex, Claude Desktop, and other MCP clients work with weather data through natural language. Ask for a forecast, compare locations, check an alert, or look up air quality — the server selects the right Meteoprog operation and returns a compact, useful answer.

Need the complete method list? See the Meteoprog API MCP Server Full Reference for all 65 smart, OpenAPI, and weather-alert tools.

Access
Natural language
Ask questions instead of choosing API endpoints.
Coverage
65 tools
Smart tools plus direct OpenAPI operations.
Safety
Read-only
The server retrieves data and changes nothing.

Start with one of the smart tools

For normal requests, you do not need to know the underlying Meteoprog endpoint. The MCP client can start with one of these tools:

`get_weather`
Current conditions and forecasts.
`compare_weather`
Compare 2–10 locations.
`get_weather_alerts`
Active severe-weather alerts.
`resolve_location`
Canonical city, coordinates, and Meteoprog ID.

Weather data for AI assistants

The server can provide current observations, date- and hour-specific forecasts, seven-day and two-week summaries, calendar-month models, weather comparisons, active alerts, alert explanations, air quality, astronomy, climate history, geomagnetic activity, water temperature, location lookup, supported languages, and service status.

Responses include units, the relevant local time period, resolved location details, data availability, and usage information. This helps an AI assistant distinguish an exact hourly forecast from a broader daily or monthly model summary.

Ask in plain language

Use MCP Meteoprog to show the current weather in Miami in English.
Use MCP Meteoprog: what will the temperature be in Warsaw tomorrow at 1:00 PM? Will it rain?
Use MCP Meteoprog to compare tomorrow's weather in Miami, New York, and Los Angeles.
Use MCP Meteoprog to check active severe-weather alerts in the United States.
Use MCP Meteoprog to show air quality in Miami for the next five days and identify the worst day.

Specify a response language — for example, “show the forecast in Spanish” — and add a country code or coordinates when a city name could be ambiguous.

Connect an MCP client

The Meteoprog MCP server supports OAuth for remote MCP clients according to the MCP Authorization specification. Add the server URL below; the client will open a browser so you can sign in and authorize access to your Meteoprog account. The OAuth flow requires your Meteoprog API key, which you enter when prompted; it does not belong in the MCP client configuration.

Free Developer Access for testing

Developer Access is a free development environment for API exploration, integration work, schema validation, and MCP testing before production. No credit card is required. It provides a real API key, 5,000 requests per day, a 5 requests per second limit, and access to the API Sandbox.

Developer Access is not a free production tier. Weather data is limited to a fixed, representative set of populated locations: Brisbane, Malaga, Oujda, Santiago, Amarillo, Cuttack, Espoo, Himeji, Samut Prakan, and Aktobe. The set covers six continents, UTC−6 to UTC+10, and representative tropical, Mediterranean, subtropical, continental, and semi-arid climate zones.

The API endpoints, schemas, response formats, and localization remain the same as production. Historical and forecast depth, location coverage, and usage limits differ. Production plans unlock more than 2,150,000 populated locations worldwide without requiring endpoint or code changes.

MCP server endpoint: https://api.meteoprog.com/mcp

Codex:

codex mcp add meteoprog --url https://api.meteoprog.com/mcp

After adding it, start an OAuth login from the Codex MCP settings and complete the browser authorization flow. Enter your Meteoprog API key when prompted.

Claude:

claude mcp add --transport http meteoprog https://api.meteoprog.com/mcp

In Claude, run /mcp, select meteoprog, and choose the authentication option to complete OAuth in your browser. Enter your Meteoprog API key when prompted; tokens are stored and refreshed by the client.

ChatGPT Desktop:

Add a new remote MCP server in the ChatGPT Desktop MCP or Connectors settings and enter https://api.meteoprog.com/mcp. Choose Connect, enter your Meteoprog API key when prompted, and complete the OAuth sign-in in your browser.

Claude Desktop:

Add a remote MCP server in Claude Desktop’s MCP settings and enter https://api.meteoprog.com/mcp. Select Connect or Authenticate, enter your Meteoprog API key when prompted, then complete the OAuth sign-in in your browser.

Keep credentials private
Enter your API key only in the secure OAuth authorization flow. Do not paste API keys or OAuth tokens into prompts, source control, or shared configuration.

Recommended smart tools

Tool Best for
get_weather Current weather and forecasts
compare_weather Comparing 2–10 locations
resolve_location Finding a canonical city and ID
get_weather_alerts Active alerts by city, coordinates, or country; optionally set {at} to evaluate alerts at a UTC minute
explain_weather_alert Explaining an alert against current conditions
get_air_quality Five-day AQI and pollutant summaries
get_astronomy Sunrise, sunset, Moon, and daylight
get_climate_history Historical extremes and climate summaries
get_geomagnetic_activity Kp and G-index observations and forecasts
get_water_temperature Water-temperature searches
get_service_status Meteoprog service health

The server also exposes low-level OpenAPI tools for exact endpoints, raw responses, geometry, historical data, and advanced integrations. Use those when direct control is required; for ordinary questions, smart tools produce clearer and smaller results.

Accuracy, usage, and expected behavior

Exact hourly forecasts are available within the next seven days. Farther dates may use a daily monthly-model summary and are marked as less precise. Monthly values are useful for broad planning and trend analysis, not exact hourly decisions.

Each upstream Meteoprog API request is accounted for separately. The MCP server does not cache weather responses, so results reflect the data available at request time. If a known city ID or coordinates are supplied, the server can avoid a separate city-name search.

Operations that could require more than 100 upstream requests require confirmation from the user. This is most relevant to large historical comparisons and bulk requests.

Forecast horizon
7 days
Exact hourly data within the supported horizon.
Locations
2–10
Locations supported in one comparison.
Server behavior
Read-only
No settings or records are changed.

Write better weather questions

  • Include the city and country when a name may be ambiguous.
  • Add a date and local hour for a precise forecast.
  • List multiple locations in one comparison request.
  • Include the month and year for a monthly forecast.
  • Specify a city, country, or coordinates for alerts. To check a past or near-future alert state, add {at} in UTC as YYYY-MM-DDTHH:mmZ (for example, 2026-08-11T14:30Z); omit it for the current state.
  • State the preferred response language.
  • Ask for a summary or table when you want a concise result.

The full MCP server exposes 13 user-oriented smart tools and 52 low-level tools generated from the Meteoprog API, for 65 tools in total.

Next step

Start building with Meteoprog Weather API

Sign up for Developer Access, get your API key instantly, and test this endpoint in API Sandbox before switching to full production access.

No credit card required
API key available instantly after signup
Sandbox requests use your API key and quota