Google Health Mcp
Give Claude Code or Cursor agents read/write access to Google Health (Fit) v4 metrics—sleep, activity, heart rate—for health, fitness, or quantified-self apps without building the API client yourself.
Overview
google-health-mcp is a Build-phase MCP server that connects AI coding agents to Google Health API v4 data—sleep, activity, and heart rate—via OAuth and stdio tools.
What is this MCP server?
- Unofficial Google Health API v4 MCP server over stdio (npm google-health-mcp-unofficial v0.4.7)
- Tools oriented to health, sleep, activity, and heart-rate data for agent-driven queries and automation
- Privacy-first positioning with optional local token path (default ~/.google-health-mcp) instead of stuffing secrets into
- google-health-mcp-server setup flow for Client ID, secret, and redirect URI stored in ~/.google-health-mcp/config.json
- OAuth via GOOGLE_HEALTH_CLIENT_ID, GOOGLE_HEALTH_CLIENT_SECRET, and GOOGLE_HEALTH_REDIRECT_URI when not using setup CLI
- Published server version 0.4.7 on npm identifier google-health-mcp-unofficial
- Transport type stdio per MCP server schema
- Four documented OAuth-related environment variables (client ID, secret, redirect URI, optional token path)
Community signal: 8 GitHub stars.
What problem does it solve?
Agents cannot see your Google Health metrics unless you maintain OAuth, token storage, and API v4 calls by hand for every health feature.
Who is it for?
Indie builders prototyping health, sleep, or activity copilots in Claude Code or Cursor who already use Google/Fit data and want MCP-native tool access.
Skip if: Teams needing verified official Google support, regulated clinical workflows, or production compliance without their own security review of an unofficial API client.
What do I get? / Deliverables
After OAuth setup, your agent can call MCP tools to fetch and reason over Google Health v4 data while tokens and optional secrets stay in local config paths you control.
- Registered stdio MCP server exposing Google Health v4-oriented tools to your agent
- Persisted OAuth tokens (default under ~/.google-health-mcp) for repeatable health, sleep, activity, and HR queries
- Optional centralized config at ~/.google-health-mcp/config.json when using google-health-mcp-server setup
Recommended MCP Servers
Journey fit
Canonical shelf is Build because the server is an external API bridge you wire into the product and agent stack during implementation, not a launch or growth tactic. Integrations fits OAuth-backed Google Health API v4 access exposed as MCP tools—classic third-party data hook for solo builders shipping agent-assisted health features.
How it compares
MCP health-data integration, not a standalone fitness app or a general-purpose agent skill.
Common Questions / FAQ
Who is google-health-mcp for?
Solo and indie developers building agent-assisted health, sleep, activity, or heart-rate features who want Google Health API v4 exposed as MCP tools instead of custom REST glue.
When should I use google-health-mcp?
Use it during Build when you are wiring integrations and want your coding agent to query or automate Google Health metrics after you complete Google Cloud OAuth.
How do I add google-health-mcp to my agent?
Install npm package google-health-mcp-unofficial (v0.4.7), run google-health-mcp-server setup or set GOOGLE_HEALTH_CLIENT_ID, GOOGLE_HEALTH_CLIENT_SECRET, and redirect URI, then add the stdio server entry to your MCP client (e.g. Claude Code or Cursor).