Arvo MCP

What khaoss85/arvo-mcp MCP server does

The khaoss85/arvo-mcp MCP server exposes Arvo fitness data and actions to AI clients that support the Model Context Protocol. It is intended for people using Arvo as an AI workout coach and want to ask questions about their training or make supported changes without switching applications.

The server includes 29 tools split between read-only and write operations. Read operations cover the user profile, training split, completed workouts, scheduled workouts, aggregated statistics, AI insights, personal records, exercise progression, exercise videos, muscle-group volume, coach details and notes, training methodology, body progress, cycle history, PT bookings, saved AI memory, caloric phases, and methodology history.

Write operations can save AI memory, change a caloric phase, update priority muscle groups, report pain or injuries, skip an exercise, generate a workout, update available equipment, add an exercise, request exercise alternatives, and apply a swap.

How it works

The khaoss85/arvo-mcp MCP server uses the standard MCP stdio transport. A compatible client starts the local Node.js process and communicates with it through MCP. The process then contacts Arvo's cloud service to retrieve or update the account data associated with the supplied API key.

Access is controlled by an Arvo API key. Arvo keys can use read-only or read-write scopes, so the selected key determines whether the client can only inspect data or also perform supported changes. The repository states that communication uses HTTPS and that the local API key is not sent to third parties.

Setup and configuration

Install and run the khaoss85/arvo-mcp MCP server with Node.js 18 or newer through the published npm package:

{
  "command": "npx",
  "args": ["-y", "arvo-mcp"],
  "env": {
    "ARVO_API_KEY": "your Arvo API key"
  }
}

Add this configuration to an MCP client such as Claude Desktop or Cursor, then restart the client. The README also identifies Windsurf and other MCP-compatible clients as supported. Create the API key from an Arvo account under Settings and API Keys. An Arvo account is required.

ARVO_API_KEY is required. ARVO_BASE_URL is optional and defaults to https://arvo.guru; it can be used when a different API base URL is needed. The server requires an internet connection because it fetches data from Arvo's cloud service.

Tools and capabilities

Typical read requests include asking for today's workout, reviewing recent PRs, checking bench-press progression, inspecting weekly muscle-group volume, or viewing body-composition changes. The tools also expose coach notes, training methodology details, bookings, and saved context for the AI.

Supported changes include modifying the current bulk, cut, or maintenance phase; updating equipment and weak points; recording a physical issue; and changing exercises in a workout. These operations can affect Arvo account data, so a read-write key should be used only when those changes are intended.

Limitations and notes

The khaoss85/arvo-mcp MCP server is not an offline tracker. Although its MCP process runs locally, it needs network access to communicate with Arvo. It also depends on an Arvo account and API key, and its available data is limited to what Arvo exposes through the listed tools and the permissions assigned to the key.

The repository documents Claude Desktop, Cursor, and Windsurf compatibility, along with standard MCP-client compatibility. It does not describe a separate web interface or local database for this server. The project is licensed under MIT.