Meteologix MCP

Python MCP server for Open-Meteo location search and Meteologix fccompact forecast data.

This package exposes Meteologix as Model Context Protocol tools:

• meteologix_search_locations
• meteologix_list_models
• meteologix_get_meteogram
• meteologix_get_current
• meteologix_get_daily_summary

Install

pip install meteologix-mcp

Run

python3 -m meteologix-mcp

MCP client configuration:

{
  "mcpServers": {
    "meteologix-weather": {
      "command": "python3",
      "args": ["-m", "meteologix-mcp"]
    }
  }
}

City Search

Use meteologix_search_locations first:

{
  "query": "shanghai",
  "limit": 5
}

Then pass the returned Open-Meteo location_id into forecast tools:

{
  "location_id": "1796236",
  "model": "rapid-euro",
  "hours": 48
}

Location search uses the Open-Meteo Geocoding API. The returned id is exposed directly as location_id, open_meteo_id, and geonames_id:

https://geocoding-api.open-meteo.com/v1/search?name=Shanghai&count=5&language=en&format=json

Forecast tools use:

https://meteologix.com/ajax_pub/fccompact

Notes

fccompact is a Meteologix web endpoint observed from the public site. It is not documented as a public stability contract by Meteologix. Meteologix currently sits behind Akamai rules that reject common non-browser clients such as Python httpx and curl, so forecast requests use curl_cffi with a browser TLS fingerprint. Use reasonable request volume and respect Meteologix terms. Location search is provided by the public Open-Meteo Geocoding API.

Maintainer Release

python -m build --no-isolation --outdir python_dist
python -m twine check python_dist/*
TWINE_USERNAME=__token__ TWINE_PASSWORD=pypi-... python -m twine upload python_dist/*