garmin-mcp

garmin-mcp

A local MCP server that exposes Garmin Connect activity, health, and training data as tools for Claude Desktop. It provides read-only access to metrics like activities, sleep, heart rate, and training readiness.

Category
访问服务器

README

garmin-mcp

A local MCP server that exposes your Garmin Connect activity, health, and training data (activities, HRV, sleep, heart rate, stress, body battery, training readiness, HR/power zones, lactate threshold, FTP, VO2max, personal records, workouts, scheduled workouts, and training plans) as tools inside Claude Desktop.

It talks directly to the live Garmin Connect API via the garminconnect library. Data is read-only — nothing is ever written back to Garmin.

How auth works

Garmin login happens in two separate places on purpose:

  • scripts/setup_auth.py — an interactive script you run yourself in a terminal. It logs in with your email/password, prompts you for an MFA code if Garmin asks for one, and saves the resulting session token to ~/.garmin_mcp/tokens.
  • The MCP server (src/garmin_mcp/server.py), launched headlessly by Claude Desktop, only ever loads and silently refreshes that cached token. It never prompts for MFA — Claude Desktop gives it no terminal to prompt on, and blocking on stdin there would hang tool calls. If the cached session is missing or has expired, tool calls fail with an error telling you to re-run setup_auth.py.

Your Garmin password is stored in the macOS keychain (item/service garmin-mcp, viewable in Keychain Access), never written to a file. Only your Garmin login email (not secret) is saved, in ~/.garmin_mcp/config.json.

Setup

  1. Install dependencies:

    cd garmin_mcp
    uv sync
    
  2. Run the one-time interactive login:

    uv run python scripts/setup_auth.py
    

    Enter your Garmin Connect email and password (the password is then stored in the keychain for next time). Enter the MFA code when prompted. On success you'll see a confirmation and the token cache path.

  3. Register the server with Claude Desktop by editing ~/Library/Application Support/Claude/claude_desktop_config.json and adding (alongside any other mcpServers entries):

    {
      "mcpServers": {
        "garmin": {
          "command": "uv",
          "args": ["run", "--project", "/Users/felix.guiony/Code/Perso/garmin_mcp", "garmin-mcp"]
        }
      }
    }
    
  4. Fully quit Claude Desktop (Cmd+Q, not just close the window) and reopen it. The tools/hammer icon should list the garmin tools.

Tools

Tool Description
list_activities Recent activities, most recent first (paginated, optional type filter)
get_activity Full detail for one activity: summary + per-metric time series
get_activity_splits Lap/split data for one activity
get_heart_rate Daily resting HR + intraday HR timeline
get_daily_stats Steps, calories, resting HR, distance, floors, intensity minutes
get_sleep Sleep stages, duration, sleep score
get_hrv Heart rate variability
get_body_battery Body Battery energy reserve over a date range
get_stress All-day stress level
get_training_readiness Training Readiness score and contributing factors
get_profile Identity, unit prefs, personal settings (max HR, resting HR, weight, height, VO2max), HR zones and power zones per sport
get_lactate_threshold Running lactate threshold: heart rate, power, and speed
get_cycling_ftp Latest cycling Functional Threshold Power
get_ftp_history Historic FTP for a sport over a date range
get_max_metrics Max-metric data (e.g. VO2max) over a date range
get_resting_heart_rate Daily resting heart rate trend over a date range
get_fitness_age Garmin Fitness Age for a given date
get_personal_records Personal records
list_workouts Saved workout templates, most recent first (paginated)
get_workout Full structure of one workout: segments, targets, intervals
get_scheduled_workouts Calendar of workouts scheduled for a given month
get_scheduled_workout Detail for one scheduled workout instance
list_training_plans Training plans
get_training_plan Details for a specific training plan
get_adaptive_training_plan Details for a specific adaptive training plan

All date parameters take YYYY-MM-DD and default to today.

Troubleshooting

A tool call returns a "Garmin session error" / "run setup_auth.py" message. Your cached refresh token expired or was revoked (this happens occasionally, e.g. after a password change or long period of inactivity). Re-run:

uv run python scripts/setup_auth.py

Rate limited. Garmin Connect will occasionally rate-limit rapid requests; the underlying library retries transient failures automatically, but if you see a rate-limit error, wait a bit and try again.

Verifying the server directly, outside Claude Desktop:

uv run mcp dev src/garmin_mcp/server.py

opens the MCP Inspector so you can call each tool by hand and check its output/schema.

Project layout

garmin_mcp/
  pyproject.toml
  src/garmin_mcp/
    auth.py      # keychain + token-cache helpers
    server.py    # FastMCP server and tool definitions
  scripts/
    setup_auth.py  # interactive one-time/occasional login

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选