使用说明

🏥 Huawei Health MCP Server

Connect your Huawei Watch health data to AI agents via Model Context Protocol

The first open-source MCP Server for Huawei Health Kit, enabling AI agents (OpenClaw, Claude Desktop, etc.) to read and analyze health data from any Huawei wearable device.

🎯 Features

30+ health data types — Heart rate, sleep, SpO2, stress, temperature, activity, ECG, and more
REST API (Cloud) — Server-side data access, no phone dependency after OAuth
MCP Protocol — Plug into any MCP-compatible AI agent
Real-time capable — Continuous heart rate, SpO2 monitoring
Privacy-first — Your data, your server, your control

📊 Supported Data

Category	Metrics
Cardiovascular	Heart rate (instant, continuous, resting, exercise), ECG, arrhythmia alerts
Sleep	Phases (deep, light, REM), naps, apnea detection, quality score
Oxygenation	Continuous SpO2, sleep SpO2
Stress	Stress level, trends, breathing exercises
Temperature	Skin temperature (day + night)
Activity	Steps, distance, calories, Activity Rings, 100+ workout modes
Women's Health	Period, ovulation, fertile window
Emotional	Mood state, emotional wellbeing (TruSense)
Advanced Sports	Golf, free diving, trail running, cycling power, running form

🏗️ Architecture

Huawei Watch GT 5 Pro
        ↓
Huawei Health (Android)
        ↓
Huawei Health Kit REST API (Cloud)
        ↓
┌─────────────────────┐
│   Backend API       │  ← OAuth, data sync, storage
│   (FastAPI)         │
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│   MCP Server        │  ← Tools for AI agents
│   (TypeScript)      │
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│   AI Agents         │  ← OpenClaw, Claude Desktop, etc.
│   (OpenClaw/LLM)    │
└─────────────────────┘

📁 Project Structure

huawei-health-mcp/
├── android-app/          # Android OAuth bridge (minimal)
│   └── app/              # Kotlin app for Health Kit authorization
├── mcp-server/           # MCP Server (TypeScript/Node.js)
│   └── src/              # Tools, types, client
├── api/                  # Backend API (Python/FastAPI)
│   └── src/              # OAuth, data sync, endpoints
├── docs/                 # Documentation
│   ├── setup.md          # Setup guide
│   └── api-reference.md  # MCP tools reference
└── .github/workflows/    # CI/CD

🚀 Quick Start

Prerequisites

Huawei Developer account (developer.huawei.com)
Health Kit enabled for your project
Node.js 18+
Python 3.11+

Backend API Setup

Clone and setup:

git clone https://github.com/sabatech-dev/huawei-health-mcp.git
cd huawei-health-mcp

Configure environment variables:

cd api
cp .env.example .env
# Edit .env with your Huawei OAuth credentials

Install dependencies and run:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn src.main:app --reload

The API will be available at http://localhost:8080 (check http://localhost:8080/api/v1/health)

Docker Deployment

# From project root
docker-compose up -d

MCP Server Setup

cd mcp-server
npm install        # use NODE_ENV=development to include devDependencies
npm run build

Configure the backend connection via environment variables:

API_URL — backend base URL (default http://localhost:8081)
API_TOKEN — bearer token if the backend enforces API_TOKEN auth

Example claude_desktop_config.json / OpenClaw MCP config:

{
  "mcpServers": {
    "huawei-health": {
      "command": "node",
      "args": ["/path/to/mcp-server/dist/index.js"],
      "env": { "API_URL": "http://localhost:8081" }
    }
  }
}

Streamable HTTP transport (optional)

Besides stdio (the default), the MCP server exposes a Streamable HTTP endpoint per the MCP spec (POST/GET/DELETE /mcp, SSE responses, sessions via the mcp-session-id header). Enable it with MCP_HTTP=1:

cd mcp-server
npm run build
MCP_HTTP=1 MCP_HTTP_PORT=3001 node dist/index.js

Point an MCP HTTP client at http://127.0.0.1:3001/mcp. Extra env vars: MCP_HTTP_HOST (default 127.0.0.1), MCP_HTTP_PORT (default 3001), LOG_LEVEL (debug|info|warn|error).

MCP Tools

Tool	Description
`get_heart_rate`	Get current and historical heart rate data
`get_sleep_data`	Get sleep analysis (phases, quality, duration)
`get_spo2`	Get blood oxygen saturation
`get_steps`	Get daily step count and distance
`get_stress`	Get stress levels and trends
`get_temperature`	Get skin temperature readings
`get_health_summary`	Get comprehensive daily/weekly health summary
`get_workout_history`	Get exercise and workout data
`get_ecg`	Get ECG analysis results
`get_auth_status`	Check Huawei API authentication status

🔐 Security

All data stays on your server
OAuth 2.0 authorization flow (authorization code + refresh, bounded token-endpoint timeouts)
Outbound Huawei API calls: rate limited, with bounded retries on 429/5xx
No third-party data storage
User consent required for each data type

📋 License

Apache-2.0 — Free and open source.

🤝 Contributing

Contributions welcome! This is a new project. See CONTRIBUTING.md for guidelines.

🙏 Credits

Built with ❤️ by SabaTech Industries — Powered by OpenClaw

✅ Current Status

✅ Backend API (FastAPI) - Fully implemented with all endpoints
✅ MCP Server (TypeScript) - All 9 tools defined, compiles successfully
✅ Docker Support - Multi-stage build, non-root runtime, healthcheck
✅ Test Suite - 60 tests passing, 86% coverage on api/src (pytest --cov)
✅ MCP Compliance - Audited against the MCP spec: see docs/mcp-compliance.md
⏳ Huawei Developer Verification - Pending (can't test real API yet)
⏳ Android OAuth Bridge - Coming soon

Note: While waiting for Huawei developer account verification, the full infrastructure is ready. Once verified, just add credentials to .env and the system will work with real Huawei Health Kit API.

Huawei Developer Project: MyHealthWatch (ID: 101653523863750125)