VitalLink MCP

What it does

VitalLink MCP turns any MCP-compatible AI client into a personal health intelligence hub. It connects to a Cloudflare Worker health data backend and exposes 10 tools covering glucose (CGM), vitals, meals, sleep, behavior, skin analysis, environmental conditions, and causal health patterns. The standout feature is render_health_insights — a tool that renders a live, interactive health dashboard widget directly inside the AI client's UI using MCP Apps.

Key features

• Live health dashboard widget rendered inside Claude via MCP Apps (@modelcontextprotocol/ext-apps) — no browser needed
• CGM glucose tracking with trend detection and energy state inference
• Causal graph exploration — see how metrics like high glycemic meals → glucose spikes → poor sleep connect to each other
• Skin analysis scores across acne, dark circles, redness, oiliness, and more
• Environmental health correlations — AQI, UV, pollen and their impact on your metrics
• Per-session isolation — each AI client session gets its own server instance

Installation

Requires Node.js 18+ and a running VitalLink Cloudflare Worker backend.

git clone https://github.com/Hrishikesh-Athreya/VitalLinkMCP.git
cd VitalLinkMCP
npm install
npm run dev
# Server starts at http://localhost:3000

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "vita-health": {
      "url": "http://localhost:3000/mcp"
    }
  }
}

Claude.ai web: Add http://localhost:3000/mcp in Settings → Integrations (requires local tunnel for remote access).

ChatGPT / OpenAI: Point a custom tool integration at http://localhost:3000/mcp using StreamableHTTP transport.

Supported hosts

Claude (Desktop and Claude.ai) and ChatGPT are explicitly documented in the README.