Connect JMA Hazard to Copilot Studio
Query JMA Hazard directly from Copilot Studio. Arlet links the two through one MCP connection and takes care of API keys.
This server connects your AI client to the hazard information published on the Japan Meteorological Agency (JMA) website (www.jma.go.jp). It answers 'is it dangerous here right now?', while the JMA Weather server answers 'what will the weather be?'.
Key features:
- get_disaster_risk — JMA's real-time landslide, inundation and flood risk (キキクル) at an exact latitude/longitude, as the five-step risk level (警戒レベル相当). Set past_minutes to include the preceding observations, so you can tell whether the risk is still rising — which matters more for an evacuation decision than the current level alone.
- get_thunder_tornado_nowcast — lightning activity (level 1-4) and tornado likelihood (level 1-2) right now and in 10-minute steps for the next hour. Use this rather than a rainfall tool when the decision is about staying outdoors.
- get_warnings — the weather warnings and advisories in force for an area, with their Japanese names (大雨警報, 雷注意報, ...), severity class and status, ordered most severe first, optionally broken down per municipality.
- get_earthquakes — recent earthquakes (roughly the last month) with epicenter, magnitude and maximum seismic intensity, and the intensity observed in each prefecture and area for one event.
- get_tsunami_information — tsunami warnings, advisories and forecasts, including the per-forecast-region breakdown that earthquake information does not carry.
- search_area — resolve a Japanese place name to the forecast area code that get_warnings takes.
Important notes:
- No API key or account is required; the data is provided free of charge.
- The risk levels and the thunder and tornado nowcast are JMA's own ordinal scales, not physical quantities. JMA publishes them as map tiles coloured by legend, not as numbers.
- The most severe risk level (災害切迫) could not be checked against live data because it did not occur anywhere in Japan while the legend was being verified; responses mark it with legend_unverified.
- Outside JMA's coverage the tiles are blank, so those locations come back as the lowest level. Every mesh response reports how many frames could not be retrieved, so 'no risk' is never confused with 'could not read'.
- For the weather forecast, rainfall amounts and AMeDAS observations, use the JMA Weather server instead.
- The data comes from JMA website content, not an officially guaranteed public API, so URLs and formats may change without notice.
- Source attribution: data from the Japan Meteorological Agency website (気象庁ホームページ). Each tool response includes the attribution.
- This server strictly relays what JMA issued and never generates its own assessment (issuing original weather forecasts is restricted under Japan's Meteorological Service Act).
- For life-threatening situations always check the JMA website, official broadcasts and your municipality's evacuation information directly; this server is not a real-time alerting system.
How to connect JMA Hazard to Copilot Studio
- 1
In the Arlet dashboard, open Data tools, find JMA Hazard and switch it on. No API key setup is needed.
- 2
Sign in to Copilot Studio with your work account, create a new agent (or open an existing one), and in the Build tab choose Tools → + → Add → Model Context Protocol (MCP) in the right panel.
- 3
In the Add MCP server form, enter "Arlet" as the Server name, https://mcp.arlet.dev/mcp/ as the Server URL, choose OAuth 2.0 for Authentication and "Dynamic (detected)" for Configuration type, then click Add. No Client ID / Secret is needed.
- 4
On the Select a connection screen, choose Create new connection → Create, then sign in with Google on the Arlet authorization screen and click Allow. Click Add to attach the tool to the agent, then Publish to use it from Microsoft 365 Copilot.
Requires a Microsoft work account and a Copilot Studio license (or trial). MCP is available only for agents using generative orchestration and not in Copilot Studio in Teams.
