JMA Data

You set up and run the JMA Data MCP Server to access real-time, historical, and forecast weather data from AMeDAS stations across Japan. This MCP server aggregates station data and provides convenient tools to query current conditions, forecasts, and past observations, enabling quick integration into your workflows and applications.

How to use

Start the MCP server using your preferred MCP client. Once the server is running, you can call the available tools to retrieve current weather, forecasts, station information, and historical data. Use the station-related tools to locate stations by code, name, or proximity, and then query weather data from those stations or nearby stations. Use the forecast tools to obtain weather outlooks by prefecture and explore historical data for trend analysis.

How to install

Prerequisites: ensure you have a runtime that can execute MCP server tasks. This server supports standard MCP client workflows and can be run either from a prebuilt executable or via development tooling mentioned below.

# Install the MCP server from the source repository
uv tool install git+https://github.com/koizumikento/jma-data-mcp.git

Or install locally by cloning the repository and syncing dependencies, then start the server via your MCP client workflow.

git clone https://github.com/koizumikento/jma-data-mcp.git
cd jma-data-mcp
uv sync

Examples

Here are typical commands you may run with your MCP client after the server is up:

# Backward-compatible MCP server startup
jma-data-mcp

# Explicit MCP server startup
jma-data-mcp serve

# Station info
jma-data-mcp station get --code 44132

# Nearby stations
jma-data-mcp station nearby --lat 35.6812 --lon 139.7671 --radius-km 10

# Current weather
jma-data-mcp weather current --station-code 44132

# Forecast
jma-data-mcp forecast get --prefecture tokyo

# Historical data
jma-data-mcp history get --station-code 44132 --target-datetime "2025-12-01 12:00"

Notes on usage and data

The server exposes tools to fetch current observations, weather forecasts, and historical data. Real-time data is typically delayed by about 30 minutes, while historical data covers the past 1–2 weeks. Time series data supports trend analysis for up to 1 week, with hourly resolution.

Configuration and MCP client setup

You can configure your MCP client to know where to reach the JMA Data MCP Server. The server can be run locally or accessed via a remote MCP runner as shown in the examples.

# Example MCP server configurations (local dev)
{
  "mcpServers": {
    "jma_data_mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/koizumikento/jma-data-mcp.git",
        "jma-data-mcp"
      ]
    }
  }
}

Usage with project cursors or integration tools

If you use tooling that reads MCP server configurations from project files, you can add the server as shown in the examples for Cursor or Cline integrations.

{
  "mcpServers": {
    "jma_data_mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/koizumikento/jma-data-mcp.git",
        "jma-data-mcp"
      ]
    }
  }
}

Available tools

get_station_info

Fetch information about a station using its code.

search_stations

Search stations by name or partial match.

search_nearby_stations

Find stations within a specified radius of given coordinates.

get_stations_of_type

Retrieve stations by type category A-F.

list_stations

List stations with pagination support.

get_current_weather

Get current weather observation from the nearest station or a specified station.

get_weather_by_location

Get weather data from the nearest station to provided coordinates.

get_forecast

Obtain weather forecast for a specific prefecture.

list_prefectures

List available prefecture codes.

get_historical_weather

Retrieve weather data for a past date/time (about 1–2 weeks available).

get_weather_time_series

Get hourly time series data for trend analysis (up to 1 week).