WHOOP MCP Server
Enables Claude to query personal WHOOP health data including recovery, sleep, strain, workouts, and profile via WHOOP's OAuth 2.0 REST API.
README
WHOOP MCP Server
An MCP (Model Context Protocol) server that lets Claude query your personal WHOOP health data — recovery, sleep, strain, workouts, and profile — via WHOOP's official OAuth 2.0 REST API (v2).
Each person who uses this runs their own copy against their own WHOOP account: you register your own free developer app with WHOOP, and your tokens/credentials stay in local files on your machine (gitignored, and written with restrictive permissions) — they're never sent anywhere except directly between your machine and WHOOP's API.
That said, the health data these tools return is a different story: once a tool call happens, its result is returned to whatever MCP client/model you're using (e.g. Claude), so it's subject to that service's own privacy and data-handling policies from that point on. Only the credentials and tokens are guaranteed to stay local — the recovery scores, sleep data, etc. you ask about necessarily become visible to the model answering your question.
git clone https://github.com/vaibhavgoel63-arch/Whoop-MCP.git
cd Whoop-MCP
1. Get WHOOP API credentials
- Go to the WHOOP Developer Dashboard and sign in.
- Create (or open) an app.
- Under the app's API settings, set the Redirect URI to exactly:
http://localhost:8080/callback - Copy the Client ID and Client Secret — you'll paste these into
.envin step 3 below.
2. Prerequisites
- Node.js 18+ (needed for the built-in
fetchAPI). Check withnode --version.
3. Install and configure
npm install
cp .env.example .env
Open .env and paste in your Client ID and Client Secret from step 1:
WHOOP_CLIENT_ID=your-client-id-here
WHOOP_CLIENT_SECRET=your-client-secret-here
WHOOP_REDIRECT_URI=http://localhost:8080/callback
⚠️ You must manually fill in
WHOOP_CLIENT_IDandWHOOP_CLIENT_SECRET— the server will refuse to start any OAuth flow until these are set.
4. Build and log in (one-time)
npm run build
npm run login
This will:
- Start a temporary local server on
http://localhost:8080. - Open your browser to WHOOP's consent screen (requesting recovery, sleep, cycle, workout, profile, body-measurement, and offline/refresh scopes).
- After you approve, WHOOP redirects back to
localhost:8080/callbackwith an authorization code. - The script exchanges that code for an access + refresh token and saves them to
token.json(gitignored) in the project root.
You only need to do this once. The MCP server automatically refreshes the access token using the
refresh token when it expires (WHOOP access tokens last about 1 hour). If your refresh token is
ever revoked or expires, just re-run npm run login.
5. Connect to Claude Desktop
Open your Claude Desktop config file:
- Windows:
%APPDATA%\Claude\claude_desktop_config.json - macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
Note: if you installed Claude Desktop from the Microsoft Store, the file above may just be a stub — the app actually reads
%LOCALAPPDATA%\Packages\<Claude package folder>\LocalCache\Roaming\Claude\claude_desktop_config.json. If tools don't show up after following the steps below, check there.
Add a whoop entry under mcpServers. Replace the path below with the absolute path to this
project's dist/server.js on your machine (find it with pwd on macOS/Linux or cd on Windows
from inside the project folder):
{
"mcpServers": {
"whoop": {
"command": "node",
"args": ["/absolute/path/to/whoop-mcp/dist/server.js"]
}
}
}
⚠️ You must manually fill in this absolute path to wherever you cloned this repo. Use forward slashes even on Windows (e.g.
C:/Users/you/whoop-mcp/dist/server.js) — they work fine inside this JSON file.
Then fully restart Claude Desktop (quit from the system tray, not just close the window).
6. Test it
- In Claude Desktop, click the tools/hammer icon in the chat box and confirm you see
whoopwith 6 tools listed:get_recovery,get_sleep,get_strain,get_workouts,get_profile,get_training_recommendation. - Try these prompts:
- "How was my recovery this week?"
- "Should I train hard today?"
- "How did I sleep the last 3 nights?"
- "Show me my workouts from the last 7 days."
- "What's my current strain and how does it compare to yesterday?"
- "What was my average strain in March 2026?" (historical range, not just "last N days")
- Spot-check one result (e.g. today's recovery score) against the WHOOP app to confirm the numbers match.
Tools reference
| Tool | Description |
|---|---|
get_recovery(days | start+end) |
Recovery score, HRV, resting heart rate, SpO2, skin temp — plus an averages summary |
get_sleep(days | start+end) |
Sleep performance %, efficiency %, stage breakdown (light/deep/REM), and sleep_onset (full ISO timestamp of when sleep began, not just the date) — plus an averages summary |
get_strain(days | start+end) |
Daily strain, average/max heart rate, calories per day — plus an averages/totals summary |
get_workouts(days | start+end) |
Logged workouts with sport, duration, strain, heart rate, calories — plus a summary |
get_profile() |
Name, email, height, weight, max heart rate |
get_training_recommendation() |
Combines today's recovery + last 3 days of sleep into a green/yellow/red training recommendation, with a sleep-debt warning if performance has been under 80% for 2+ nights |
The four range-based tools accept either days (rolling window, e.g. days=7 for the last week) or
an explicit start/end date pair (YYYY-MM-DD, end exclusive) for querying a specific historical
period, e.g. start="2026-03-01", end="2026-04-01" for all of March 2026. Each returns a summary
object (averages/totals) alongside the individual daily/nightly records.
Troubleshooting
- "No WHOOP tokens found" — run
npm run login. - 401 / token errors after working before — the server auto-refreshes access tokens; if you see
a refresh failure, your refresh token was likely revoked (e.g. you removed app access in WHOOP
settings). Re-run
npm run login. - 403 Forbidden — your token is missing a scope. Scopes are fixed at login time, so re-run
npm run loginto get a fresh token with the full scope set. - 429 Too Many Requests — you've hit WHOOP's rate limit (100 requests/minute, 10,000/day). Wait and try again.
- Tools don't show up in Claude Desktop — double-check the absolute path in
claude_desktop_config.json, that you rannpm run build(the config points atdist/server.js, notsrc/server.ts), and that you fully restarted Claude Desktop.
Project structure
src/
auth.ts # OAuth constants, token load/save, refresh logic
login.ts # One-time login script (npm run login)
whoopClient.ts # Authenticated WHOOP API client + response normalizers
server.ts # MCP server exposing the 6 tools
.env.example # Template for WHOOP_CLIENT_ID / WHOOP_CLIENT_SECRET / WHOOP_REDIRECT_URI
.env (credentials) and token.json (access/refresh tokens) are both gitignored — never commit
either file.
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
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.