Japan Intel MCP

What japan-intel-mcp MCP server does

The japan-intel-mcp MCP server exposes eight read-only operations for Japanese public information. Its coverage includes recent earthquakes, active weather warnings and advisories, narrative weather overviews, seven-day forecasts, active typhoons, statutory corporate disclosures, postal-code lookups, and a combined watch summary.

The earthquake operation can filter results by minimum magnitude and seismic intensity, then return details such as epicenter, depth, magnitude, and maximum intensity. Weather warning results can be narrowed by prefecture. Forecast data includes high and low temperatures, precipitation probability, and reliability information. The weather overview is available for ten major prefectures.

Corporate filing searches use EDINET data and cover disclosures such as annual reports and timely disclosure. Postal lookup converts a Japanese postal code into an address in kanji and kana. The watch summary provides a single call for a cross-source view, with a minimum-magnitude filter for earthquake data.

How it works

The server gathers data from the Japan Meteorological Agency (JMA), EDINET, and zipcloud, then presents those sources through MCP tools. It is designed for live public-data retrieval rather than local document search or data mutation. No API key, account, or paid API tier is required according to the project documentation.

Each response follows a common structure containing a retrieval timestamp, a sources list, data_gaps, and the returned data. The source list lets an agent identify where the result came from. If a source could not be retrieved, the gap is reported instead of being silently excluded. This makes the response suitable for workflows that need to distinguish an empty result from incomplete retrieval.

Setup and configuration

Install the package from a checked-out repository with:

pip install -e .

Start the japan-intel-mcp MCP server in stdio mode with either japan-intel-mcp or python -m japan_intel_mcp.server. The documented stdio configuration can be used by Claude Code and Cursor, and the repository also shows a command-based configuration for Hermes. A versioned MCP bundle is available from the project’s v0.2.1 release, and the project is listed in the official MCP registry. The documentation states that no environment variables or API keys are needed.

Example calls include filtering earthquakes by magnitude, requesting Tokyo warnings or forecasts, looking up postal code 1000001, and generating a watch summary with a minimum magnitude of 3.0.

Tools and capabilities

• japan_earthquakes: Retrieve recent earthquake events with magnitude, intensity, depth, and epicenter information.
• japan_weather_warnings: Get active warnings and advisories, optionally filtered by area or prefecture.
• japan_weather_overview: Request a narrative weather summary for a supported major prefecture.
• japan_weekly_forecast: Retrieve seven-day temperature, precipitation-probability, and reliability data.
• japan_typhoons: Check active typhoon tracking information.
• japan_edinet_filings: Find statutory corporate filings for a specified date.
• japan_postal_lookup: Resolve a postal code to kanji and kana address forms.
• japan_watch_summary: Combine relevant cross-source information in one request.

Limitations and notes

The tools are read-only; the material does not describe create, update, or delete operations. Weather overviews cover ten major prefectures rather than every location. Results depend on the availability of the upstream public sources, and any retrieval shortfall is surfaced in data_gaps. The project documentation identifies the MCP bundle as v0.2.1 and describes PyPI publication as in progress, so installation options may differ by release channel.