A Model Context Protocol (MCP) server implementation that interfaces with the Hevy fitness tracking app and its API. This server enables AI assistants to access and manage workout data, routines, exercise templates, and more through the Hevy API (requires PRO subscription).

• Workout Management: Fetch, create, and update workouts
• Routine Management: Access and manage workout routines
• Exercise Templates: Browse available exercise templates
• Folder Organization: Manage routine folders
• Webhook Subscriptions: Create, view, and delete webhook subscriptions for workout events

Note: HTTP transport and Docker images remain deprecated. Smithery deployment now uses the official TypeScript runtime flow (no Docker required), or you can run the server locally via stdio (e.g., npx hevy-mcp). Existing GHCR images remain available but are no longer updated.

Pick the workflow that fits your setup:

• Node.js (v20 or higher; strongly recommended to use the exact version pinned in .nvmrc to match CI)
• pnpm (via Corepack)
• A Hevy API key
  • Optional: A Smithery account + API key/login if you plan to deploy via Smithery

You can launch the server directly without cloning:

HEVY_API_KEY=your_hevy_api_key_here npx -y hevy-mcp

# Clone the repository
git clone https://github.com/chrisdoc/hevy-mcp.git
cd hevy-mcp

# Install dependencies
corepack use [email protected]
pnpm install

# Create .env and add your keys (never commit real keys)
cp .env.sample .env
# Edit .env and add at least HEVY_API_KEY. Add SMITHERY_API_KEY if you use Smithery CLI.

To use this MCP server with Cursor, add/merge this server entry under "mcpServers" in ~/.cursor/mcp.json:

{
  "hevy-mcp": {
    "command": "npx",
    "args": ["-y", "hevy-mcp"],
    "env": {
      "HEVY_API_KEY": "your-api-key-here"
    }
  }
}

Make sure to replace your-api-key-here with your actual Hevy API key.

If your mcp.json already contains other servers, do not replace the whole file—merge the "hevy-mcp" entry into your existing "mcpServers" object.

The "hevy-mcp" key name is arbitrary. If you already have an existing config using a different name (for example "hevy-mcp-server"), you can keep it.

If you already have an existing "mcpServers" object, merge the "hevy-mcp" entry into it without removing other servers.

{
  "mcpServers": {
    "hevy-mcp": {
      "command": "npx",
      "args": ["-y", "hevy-mcp"],
      "env": {
        "HEVY_API_KEY": "your-api-key-here"
      }
    }
  }
}

You can supply your Hevy API key in two ways:

1. Environment variable (HEVY_API_KEY)
2. Command-line argument (--hevy-api-key=your_key or hevy-api-key=your_key after -- when using pnpm scripts)

Create a .env file in the project root (you can copy from .env.sample) with the following content if using the environment variable approach:

HEVY_API_KEY=your_hevy_api_key_here

Replace your_hevy_api_key_here with your actual Hevy API key. If you prefer the command argument approach you can skip setting the environment variable and start the server with for example:

pnpm start -- --hevy-api-key=your_hevy_api_key_here

hevy-mcp ships with Sentry monitoring baked into the built MCP server so that usage and errors from published builds can be observed.

The server initializes @sentry/node with a fixed DSN, release name derived from the package version, and tracing settings directly in the code (see src/index.ts), and wraps the underlying McpServer with Sentry.wrapMcpServerWithSentry so requests and tool calls are captured by Sentry automatically. The configuration uses sendDefaultPii: false to keep Sentry's default PII collection disabled.

There is currently no built-in toggle to disable Sentry for the published package. If you need a build without Sentry telemetry, you can fork the repository and remove the Sentry initialization in src/index.ts.

Smithery can bundle and host hevy-mcp without Docker by importing the exported createServer and configSchema from src/index.ts.

1. Ensure dependencies are installed: pnpm install

2. Launch the Smithery playground locally:

   pnpm run smithery:dev

   The CLI will prompt for HEVY_API_KEY, invoke createServer({ config }), and open the Smithery MCP playground.

3. Build the deployable bundle:

   pnpm run smithery:build

4. Connect the repository to Smithery and trigger a deployment from their dashboard. Configuration is handled entirely through the exported Zod schema, so no additional smithery.yaml env mapping is required.

Why are chalk, cors, and @smithery/sdk dependencies? Smithery’s TypeScript runtime injects its own Express bootstrap that imports these packages. Declaring them in package.json ensures the Smithery CLI can bundle your server successfully.

As of version 1.18.0, hevy-mcp only supports stdio transport. HTTP/SSE transport has been completely removed.

hevy-mcp runs exclusively over stdio, which works seamlessly with MCP-aware clients like Claude Desktop and Cursor. The server communicates via standard input/output streams using JSON-RPC messages.

If you were using HTTP or SSE transport in an older version (< 1.18.0), you must migrate to stdio.

The HTTP/SSE transport was removed in v1.18.0 to simplify the codebase and focus on the stdio-native MCP experience. If you're encountering errors like:

• "stream is not readable" when making HTTP requests
• "HTTP transport mode has been removed from hevy-mcp"
• Server messages about SSE mode on http://localhost:3001

You are likely running an outdated build or trying to connect with an HTTP-based client. Here's how to fix it:

1. Update to the latest version:

   npx -y hevy-mcp@latest
   # or if installed locally:
   pnpm install hevy-mcp@latest

2. Update your client configuration to use stdio transport instead of HTTP. For example, in Cursor's ~/.cursor/mcp.json:

   Old HTTP-based config (no longer supported):

   {
     "hevy-mcp": {
       "url": "http://localhost:3001/sse"
     }
   }

   New stdio-based config (current):

   {
     "hevy-mcp": {
       "command": "npx",
       "args": ["-y", "hevy-mcp"],
       "env": {
         "HEVY_API_KEY": "your-api-key-here"
       }
     }
   }

3. Clear any cached builds:

   # If you have a local clone, rebuild
   pnpm run build
   
   # Or remove node_modules and reinstall
   rm -rf node_modules dist
   pnpm install
   pnpm run build

4. Ensure you're not running a custom HTTP server. If you have custom code that imports createHttpServer(), it will now throw an error. Remove those imports and use stdio transport instead.

If you absolutely need HTTP/SSE transport, you can use version 1.17.x or earlier, but those versions are no longer maintained and contain known bugs (including the "stream is not readable" issue caused by middleware conflicts).

pnpm run dev

This starts the MCP server in development mode with hot reloading.

pnpm run build
pnpm start

Docker-based workflows have been retired so we can focus on the stdio-native experience. The bundled Dockerfile now exits with a clear message to prevent accidental builds, and .dockerignore simply documents the deprecation. Previously published images remain available on GHCR (for example ghcr.io/chrisdoc/hevy-mcp:latest), but they are no longer updated. For the best experience, run the server locally via npx hevy-mcp or your own Node.js runtime.

The server implements the following MCP tools for interacting with the Hevy API:

• get-workouts: Fetch and format workout data
• get-workout: Get a single workout by ID
• create-workout: Create a new workout
• update-workout: Update an existing workout
• get-workout-count: Get the total count of workouts
• get-workout-events: Get workout update/delete events

• get-routines: Fetch and format routine data
• create-routine: Create a new routine
• update-routine: Update an existing routine
• get-routine-by-id: Get a single routine by ID using direct endpoint

• get-exercise-templates: Fetch exercise templates
• get-exercise-template: Get a template by ID

• get-routine-folders: Fetch routine folders
• create-routine-folder: Create a new folder
• get-routine-folder: Get a folder by ID

• get-webhook-subscription: Get the current webhook subscription
• create-webhook-subscription: Create a new webhook subscription
• delete-webhook-subscription: Delete the current webhook subscription

hevy-mcp/
├── .env                   # Environment variables (API keys)
├── src/
│   ├── index.ts           # Main entry point
│   ├── tools/             # Directory for MCP tool implementations
│   │   ├── workouts.ts    # Workout-related tools
│   │   ├── routines.ts    # Routine-related tools
│   │   ├── templates.ts   # Exercise template tools
│   │   ├── folders.ts     # Routine folder tools
│   │   └── webhooks.ts    # Webhook subscription tools
│   ├── generated/         # API client (generated code)
│   │   ├── client/        # Kubb-generated client
│   │   │   ├── api/       # API client methods
│   │   │   ├── types/     # TypeScript types
│   │   │   ├── schemas/   # Zod schemas
│   │   │   └── mocks/     # Mock data
│   └── utils/             # Helper utilities
│       ├── config.ts              # Env/CLI config parsing
│       ├── error-handler.ts       # Tool error wrapper + response builder
│       ├── formatters.ts          # Domain formatting helpers
│       ├── hevyClient.ts          # API client factory
│       ├── httpServer.ts          # Legacy HTTP transport (deprecated; throws explicit error; kept only for backward compatibility - removing may be breaking)
│       ├── response-formatter.ts  # MCP response utilities
│       └── tool-helpers.ts        # Zod schema -> TS type inference
├── scripts/               # Build and utility scripts
└── tests/                 # Test suite
    ├── integration/       # Integration tests with real API
    │   └── hevy-mcp.integration.test.ts  # MCP server integration tests

This project uses Biome for code formatting and linting:

pnpm run check

To run all tests (unit and integration), use:

pnpm test

Note: pnpm test runs all tests. Integration tests will fail by design if HEVY_API_KEY is missing. If you don’t have an API key locally, use the unit test command below.

To run only unit tests (excluding integration tests):

pnpm vitest run --exclude tests/integration/**

Or with coverage:

pnpm vitest run --coverage --exclude tests/integration/**

To run only the integration tests (requires a valid HEVY_API_KEY):

pnpm vitest run tests/integration

Note: The integration tests will fail if the HEVY_API_KEY environment variable is not set. This is by design to ensure that the tests are always run with a valid API key.

For GitHub Actions:

1. Unit + integration tests are executed as part of the normal Build and Test workflow
2. Integration tests require the HEVY_API_KEY secret to be set

The workflow runs pnpm vitest run --coverage and provides HEVY_API_KEY from repository secrets.

To set up the HEVY_API_KEY secret:

1. Go to your GitHub repository
2. Click on "Settings" > "Secrets and variables" > "Actions"
3. Click on "New repository secret"
4. Set the name to HEVY_API_KEY and the value to your Hevy API key
5. Click "Add secret"

If the secret is not set, the integration tests will fail (by design).

To set up Sentry secrets for source map uploads during builds:

The build process uses Sentry's Rollup plugin to upload source maps. You need to configure three secrets:

1. Go to your GitHub repository

2. Click on "Settings" > "Secrets and variables" > "Actions"

3. Add the following secrets:

   • SENTRY_ORG: Your Sentry organization slug
   • SENTRY_PROJECT: Your Sentry project slug
   • SENTRY_AUTH_TOKEN: A Sentry auth token with project:releases scope

   You can create a Sentry auth token at: https://sentry.io/settings/account/api/auth-tokens/

If these secrets are not set, the build will still succeed, but source maps will not be uploaded to Sentry.

Note: GitHub does not provide secrets to pull requests from forks by default, so fork PRs may fail CI unless a maintainer reruns the checks with HEVY_API_KEY available.

If CI is failing only because the fork PR is missing HEVY_API_KEY, that is expected; maintainers may rerun the workflow with secrets enabled.

For contributors from forks: CI failures caused solely by missing HEVY_API_KEY do not indicate a problem with your changes.

All other CI checks (build, formatting/linting, unit tests, etc.) are still expected to pass.

Only failures caused solely by missing HEVY_API_KEY on forked PRs are considered acceptable.

The API client is generated from the OpenAPI specification using Kubb:

pnpm run export-specs
pnpm run build:client

Kubb generates TypeScript types, API clients, Zod schemas, and mock data from the OpenAPI specification.

• Rollup optional dependency missing: If you see an error similar to Cannot find module @rollup/rollup-linux-x64-gnu, set the environment variable ROLLUP_SKIP_NODEJS_NATIVE_BUILD=true before running pnpm run build. This forces Rollup to use the pure JavaScript fallback and avoids the npm optional dependency bug on some Linux runners.

• smithery.yaml validation failed (unexpected fields): Only runtime, target, and env are allowed for the TypeScript runtime. Remove entry, name, or other fields.
• Could not resolve "chalk"/"cors": Run pnpm install so the runtime dependencies listed in package.json are present before invoking Smithery.
• Failed to connect to Smithery API: Unauthorized: Log in via pnpm dlx @smithery/cli login or set SMITHERY_API_KEY in .env.
• Tunnel crashes with RangeError: Invalid count value: This is a known issue in certain @smithery/cli builds. Upgrade/downgrade the CLI (e.g., pnpm add -D @smithery/cli@latest) or contact Smithery support.

This project is licensed under the MIT License - see the LICENSE file for details.

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

• Model Context Protocol for the MCP SDK
• Hevy for their fitness tracking platform and API