LibreLink MCP Server
Enables Claude Desktop to securely access FreeStyle LibreLink continuous glucose monitoring data, providing real-time readings, historical analysis, and health insights through natural language queries.
README
LibreLink MCP Server
A local Model Context Protocol (MCP) server that provides Claude Desktop with secure access to your FreeStyle LibreLink continuous glucose monitoring (CGM) data.
🌟 Features
- Real-time glucose monitoring - Get current readings with trend arrows
- Historical data analysis - Retrieve glucose history over customizable periods
- Comprehensive analytics - Time-in-range, GMI, variability metrics
- Pattern recognition - Dawn phenomenon, meal responses, stability analysis
- Privacy-first design - All data stays local on your machine
- Secure credential management - Local encrypted storage
- Cross-platform health integration - Works alongside other health MCP servers
📋 Prerequisites
- LibreLink Account: Active FreeStyle LibreLink account with glucose data
- Compatible Sensor: FreeStyle Libre 2 or 3 with data sharing enabled
- Node.js: Version 18.0.0 or higher
- Claude Desktop: For MCP integration
🚀 Quick Start
1. Installation
# Clone the repository
git clone https://github.com/yourusername/librelink-mcp-server.git
cd librelink-mcp-server
# Install dependencies
npm install
# Build the project
npm run build
2. Configuration
# Configure your LibreLink credentials
npm run configure
You'll be prompted for:
- Email: Your LibreLink account email
- Password: Your LibreLink account password
- Region: US or EU (based on your location)
- Target ranges: Glucose target ranges (default: 70-180 mg/dL)
3. Test Connection
# Test your LibreLink connection
node test-real-connection.js
4. Claude Desktop Integration
Add to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
{
"mcpServers": {
"librelink": {
"command": "node",
"args": ["/path/to/librelink-mcp-server/dist/index.js"]
}
}
}
5. Restart Claude Desktop
Restart Claude Desktop to load the new MCP server.
🩸 Usage Examples
Once integrated with Claude Desktop, you can ask:
Basic Glucose Queries
- "What's my current glucose level?"
- "Show me my glucose readings from the past 6 hours"
- "What's my average glucose today?"
Analytics & Insights
- "Calculate my time in range for this week"
- "Analyze my glucose patterns and trends"
- "Do I have dawn phenomenon?"
- "How stable are my overnight glucose levels?"
Health Correlations
When combined with other health MCP servers:
- "How does my sleep quality affect my glucose control?"
- "Compare my glucose variability with my stress levels"
- "Show the impact of my supplements on glucose stability"
🛠 Available MCP Tools
| Tool | Description | Parameters |
|---|---|---|
get_current_glucose |
Real-time glucose reading with trend | None |
get_glucose_history |
Historical glucose data | hours (default: 24) |
get_glucose_stats |
Statistics and time-in-range | days (default: 7) |
get_glucose_trends |
Pattern analysis | period (daily/weekly/monthly) |
get_sensor_info |
Sensor status and info | None |
configure_credentials |
Update LibreLink credentials | email, password, region |
configure_ranges |
Set target glucose ranges | target_low, target_high |
validate_connection |
Test LibreLink connection | None |
📊 Sample Output
Current Glucose Reading
{
"current_glucose": 105,
"timestamp": "2025-07-14T21:19:24.000Z",
"trend": "Flat",
"status": "Normal",
"color": "green"
}
Glucose Statistics
{
"analysis_period_days": 7,
"average_glucose": 93.46,
"glucose_management_indicator": 5.55,
"time_in_range": {
"target_70_180": 100.0,
"below_70": 0.0,
"above_180": 0.0
},
"variability": {
"standard_deviation": 7.52,
"coefficient_of_variation": 8.04
}
}
Trend Analysis
{
"period": "daily",
"patterns": [
"Good postprandial glucose control",
"Excellent overnight glucose stability"
],
"dawn_phenomenon": false,
"meal_response_average": 0,
"overnight_stability": 2.08
}
🔧 Development
Running Tests
# Run all tests
npm test
# Test MCP protocol
npm run test:mcp
# Test analytics with mock data
npm run test:analytics
# Test with real LibreLink data (requires configuration)
node test-real-data.js
Building
# Build TypeScript
npm run build
# Type checking
npm run typecheck
# Development mode
npm run dev
Project Structure
librelink-mcp-server/
├── src/
│ ├── index.ts # Main MCP server
│ ├── librelink-client.ts # LibreLink API wrapper
│ ├── glucose-analytics.ts # Analytics and statistics
│ ├── config.ts # Configuration management
│ ├── configure.ts # CLI configuration tool
│ └── types.ts # TypeScript definitions
├── config/
│ └── default.json # Default configuration
├── test-*.js # Test suites
├── package.json
└── README.md
🔒 Security & Privacy
Data Privacy
- Local processing only - No data sent to external servers
- Your data stays on your machine - Complete privacy control
- No analytics or tracking - Zero telemetry
Credential Security
- Local storage - Credentials stored in
~/.librelink-mcp/config.json - File permissions - Automatically set to user-only access (600)
- No cloud storage - Never uploaded or shared
Security Best Practices
# Verify file permissions
ls -la ~/.librelink-mcp/config.json
# Should show: -rw------- (user read/write only)
# Optional: Encrypt config directory
# (Implementation details in documentation)
⚠️ Important Notes
LibreLink API Usage
- This project uses an unofficial API through reverse engineering
- Not affiliated with Abbott or FreeStyle Libre
- Use at your own discretion and ensure compliance with LibreLink terms
- API may change - community maintained compatibility
Data Sharing Requirements
- Ensure your LibreLink app has data sharing enabled
- Your sensor must be active and transmitting data
- LibreLink account (not LibreLinkUp) credentials required
Sensor Compatibility
- ✅ FreeStyle Libre 2
- ✅ FreeStyle Libre 3
- ❓ FreeStyle Libre 1 (may work, not tested)
🐛 Troubleshooting
Common Issues
"No connections found"
- Verify you're using LibreLink (not LibreLinkUp) credentials
- Check that data sharing is enabled in your LibreLink app
- Ensure your sensor is active and connected
"Authentication failed"
- Double-check email and password
- Verify correct region (US/EU)
- Try logging into LibreLink app to confirm credentials
"Connection timeout"
- Check internet connection
- Verify LibreLink service status
- Try again after a few minutes
Getting Help
-
Run diagnostics:
node diagnose-account.js -
Check logs: Look for error messages in the console output
-
Test connection:
node test-real-connection.js -
Open an issue: Include diagnostic output and error messages
🤝 Contributing
We welcome contributions! Please:
- Fork the repository
- Create a feature branch:
git checkout -b feature/amazing-feature - Make your changes with tests
- Follow the existing code style
- Submit a pull request
Development Guidelines
- TypeScript required - Maintain type safety
- Test coverage - Add tests for new features
- Documentation - Update README for new functionality
- Security first - Never commit credentials or sensitive data
📜 License
This project is licensed under the MIT License - see the LICENSE file for details.
🙏 Acknowledgments
- libre-link-unofficial-api - Community-maintained LibreLink API client
- MCP Protocol - Anthropic's Model Context Protocol
- FreeStyle Libre Community - Inspiration and reverse engineering efforts
- Open Source Diabetes Projects - Nightscout, OpenAPS, and others
⭐ Support
If this project helps you manage your diabetes with AI assistance, please:
- ⭐ Star the repository
- 🐛 Report issues you encounter
- 💡 Suggest improvements
- 🤝 Contribute to the project
Disclaimer: This is an unofficial project not affiliated with Abbott or FreeStyle Libre. Use responsibly and in compliance with applicable terms of service. Always consult healthcare professionals for medical decisions.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。