Tokyo WBGT MCP Server

# CLAUDE.md This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository. ## Project Overview **Tokyo WBGT MCP Server** - A Model Context Protocol (MCP) server that provides access to Tokyo's WBGT (Wet Bulb Globe Temperature) forecast and real-time data from Japan's Ministry of the Environment. WBGT is a heat index used for heatstroke prevention. ## Key Development Commands ```bash # Install dependencies npm install # Build TypeScript to JavaScript npm run build # Run in production (requires build first) npm start # Run in development mode npm run dev ``` ## Architecture Overview ### Single Class Design The entire server is implemented as a single `TokyoWBGTServer` class in `tokyo_wbgt_mcp_server.ts` that: - Implements MCP SDK patterns with request handlers - Exposes 3 tools: forecast data, all locations, and real-time data - Fetches and parses CSV data from Ministry of Environment APIs - Provides structured JSON responses with WBGT risk levels ### Data Flow 1. MCP client calls tool via JSON-RPC → 2. Server routes to handler method → 3. Fetches CSV from `wbgt.env.go.jp` → 4. Parses CSV to structured JSON → 5. Returns formatted response ### Core Methods - **Data Fetchers**: `getTokyoWBGTForecast()`, `getAllTokyoLocations()`, `getTokyoRealtimeData()` - **CSV Parsers**: `parseWBGTForecastCSV()`, `parseAllTokyoCSV()`, `parseRealtimeCSV()` - **Utilities**: `formatDateTime()`, `getWBGTStatus()`, location mapping functions ## Project Structure ### Source Code Organization - **src/index.ts**: Main MCP server class - **src/types.ts**: TypeScript type definitions - **src/constants.ts**: Location mappings and API URLs - **src/utils.ts**: CSV parsing and utility functions - **src/wbgt-service.ts**: Data fetching service class ### Build Output - All TypeScript files compile to `build/` directory - Entry point is `build/index.js` ## Tokyo Location Data The server supports 7 Tokyo observation stations with specific station IDs: - Tokyo (Koishikawa): 44132 (real-time available) - Hachioji: 44112, Fuchu: 44116, Nerima: 44071 - Ome: 44056, Ogochi: 44046, Edogawa Rinkai: 44136 ## WBGT Risk Levels - Safe (< 21°C), Caution (21-25°C), Alert (25-28°C), High Alert (28-31°C), Danger (≥ 31°C) ## External Dependencies - Fetches data from `https://www.wbgt.env.go.jp/` APIs - No rate limiting implemented - Data updates hourly