WHOOP MCP Server
A multi-user MCP server that exposes WHOOP fitness data (recovery, sleep, workouts, cycles, profile, body measurements) to AI clients. Each user connects their own WHOOP account via OAuth and only sees their own data, which is fetched live on every tool call.
README
WHOOP MCP Server
A multi-user MCP server that exposes WHOOP fitness data (recovery, sleep, workouts, cycles, profile, body measurements) to AI clients such as Claude, ChatGPT, and local editors (Cursor, Cline).
Each user connects their own WHOOP account via OAuth and only ever sees their own data. The server acts as a full OAuth 2.1 authorization server for MCP clients, delegating identity to WHOOP as the upstream identity provider.
Data is fetched live from the WHOOP API on every tool call — no fitness data is cached or stored. Postgres only holds user identity and encrypted tokens.
Two ways to use this
There are two ways to get WHOOP data into your AI client:
- A. Self-host as a remote connector (this README) — run your own instance (Railway 1-click, Docker, Fly.io, Render, ...) and connect any MCP client to it over HTTP, either via OAuth or a personal access token.
- B. Local Claude Desktop extension (see Local Claude Desktop
extension below) — a free, packaged
.mcpbextension that runs entirely on your machine. No hosting, no database: your WHOOP tokens are encrypted and stored locally under your OS config directory. Claude Desktop only (not the browser).
1-Click deploy on Railway
- Click the button above — Railway provisions the app service and a Postgres
database, and auto-generates
ENCRYPTION_KEYandSESSION_SECRET. - Copy your new Railway domain.
- Create a WHOOP developer app (see WHOOP app setup below).
- Enter
WHOOP_CLIENT_IDandWHOOP_CLIENT_SECRETas service variables in Railway, then redeploy. - Connect an AI client (see Connect an AI client below).
WHOOP app setup
-
Register a developer application at the WHOOP Developer Portal.
-
Set the app's redirect URI to:
https://<your-domain>/whoop/callback -
Request the scopes:
read:recovery read:sleep read:workout read:cycles read:profile read:body_measurement offline -
Copy the client ID and secret — you'll enter these as
WHOOP_CLIENT_ID/WHOOP_CLIENT_SECRET.
Running on other hosts
Copy .env.example to .env and fill in:
| Variable | Description |
|---|---|
DATABASE_URL |
Postgres connection string. |
ENCRYPTION_KEY |
32-byte hex key used to encrypt stored WHOOP tokens at rest. Generate with openssl rand -hex 32. |
WHOOP_CLIENT_ID |
Client ID from your WHOOP developer app. |
WHOOP_CLIENT_SECRET |
Client secret from your WHOOP developer app. |
PUBLIC_BASE_URL |
The publicly reachable base URL of this deployment, e.g. https://your-app.example.com (no trailing slash). Used to build OAuth redirect/callback URLs. On Railway this is derived automatically from RAILWAY_PUBLIC_DOMAIN when not set explicitly; on other hosts you must set it explicitly. |
SESSION_SECRET |
Secret used to sign the dashboard's session cookie. Generate with openssl rand -hex 32. |
PORT |
Port to listen on. Defaults to 8080. |
Docker / your own VPS
docker build -t whoop-mcp . && docker run -p 8080:8080 --env-file .env whoop-mcp
Set PUBLIC_BASE_URL explicitly in your .env — there's no
RAILWAY_PUBLIC_DOMAIN to fall back on outside Railway.
Fly.io
fly launch uses the repo's Dockerfile. Set secrets via
fly secrets set KEY=value ..., and provision Postgres via fly postgres create.
Render
Create a Web Service from this repo using the Docker environment, add a Postgres add-on, and set the env vars from the table above in the Render dashboard.
Connect an AI client
Browser (Claude.ai / ChatGPT)
-
Add a remote MCP connector pointing at:
https://<your-domain>/mcp -
The client discovers the OAuth endpoints automatically, redirects you through "Connect WHOOP", and starts calling tools (e.g.
get_daily_summary) under your own WHOOP identity.
Claude Desktop (remote)
Add a custom connector with the same URL, https://<your-domain>/mcp, and
go through the same "Connect WHOOP" OAuth flow.
Optional: local clients via personal access token
- Open the dashboard at
https://<your-domain>/dashboardand click Connect WHOOP. - Once connected, click Create personal access token and copy the token (it is only shown once).
- Configure your client's
mcp.jsonto callhttps://<your-domain>/mcpwith the token as a bearer header, for example:
{
"mcpServers": {
"whoop": {
"url": "https://<your-domain>/mcp",
"headers": {
"Authorization": "Bearer <PAT>"
}
}
}
}
Revoke a token at any time from the dashboard, or delete your account entirely (removes your WHOOP connection and all issued tokens).
Sharing your instance with friends (multi-user)
The server is multi-user by design: anyone you point at your deployment
connects their own WHOOP account and only ever sees their own data —
exactly how the project author runs it for a small circle. To invite a friend,
just share your https://<your-domain>/mcp connector URL (or the dashboard
link); they go through their own "Connect WHOOP" flow. There's no per-user
setup on your side.
Two things to keep in mind before inviting people:
- WHOOP app user limit: a WHOOP developer app in development mode is capped at a small number of users (around 10). Fine for friends; a larger audience requires WHOOP's production review.
- Data responsibility: once others connect, their (encrypted) WHOOP tokens live in your database, and you become responsible for that data (e.g. GDPR if you're in the EU). Only invite people who are comfortable with that.
Local Claude Desktop extension (Way B)
A free, local alternative to self-hosting: a .mcpb extension that runs as a
subprocess inside Claude Desktop. There's no server to host and no database —
your WHOOP tokens are encrypted and stored on disk under your OS config
directory. Claude Desktop only; this doesn't work in the browser (Claude.ai
or ChatGPT), since those need a remote connector.
1. Get the extension
Either build it yourself:
npm run bundle:local
which produces whoop-mcp.mcpb in the repo root, or download the latest
whoop-mcp.mcpb from the project's releases.
2. Install it
Install the .mcpb into Claude Desktop with any of these — if a double-click
does nothing (macOS may not associate the .mcpb file type), use the
Advanced settings route, which always works:
- Advanced settings (most reliable): in Claude Desktop go to Settings →
Extensions → Advanced settings → Extension Developer → Install Extension…
and pick your
whoop-mcp.mcpb. - Drag & drop: drag
whoop-mcp.mcpbonto the Claude Desktop Settings window. - Double-click: open
whoop-mcp.mcpbfrom Finder (works on most, but not all, installs).
The install dialog asks for your WHOOP Client ID and WHOOP Client Secret — enter the credentials from your own WHOOP developer app (see below). The secret is stored securely in your OS keychain.
3. Create your own WHOOP app
-
Register a developer application at the WHOOP Developer Portal.
-
Set the app's redirect URI to exactly:
https://caxtmann.github.io/whoop-mcp/callback/This callback page is a tiny static page — it only displays the
codeandstatefor you to copy back, holds no secret, and stores no data. It is shared by all users of this extension. If you'd rather not depend on it, host your own copy ofcallback/index.html(e.g. on your own GitHub Pages), use that URL as the redirect URI here, and set the optional Redirect URI field in the install dialog to the same URL — it overrides the default. -
Request the scopes:
read:recovery read:sleep read:workout read:cycles read:profile read:body_measurement offline -
Copy the client ID and secret — enter these in the install dialog above.
4. Log in
- Run the
whoop_logintool. - Open the URL it returns and approve access on WHOOP.
- WHOOP redirects to the callback page, which displays a
codeandstate— copy both. - Run the
whoop_complete_logintool with thatcodeandstate.
5. Use it
Once logged in, call the data tools directly, e.g. get_daily_summary,
get_recovery, get_sleep.
Running locally (development)
- Set up a local Postgres database and configure
DATABASE_URL. - Run the setup commands:
npm install
npm run db:migrate
npm run dev
The server will start at http://localhost:8080. Check health with GET /healthz, which returns {"ok":true} once running.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.