Ask Claude or ChatGPT about your WHOOP data. Recovery, sleep, strain, and workout analysis computed from your own records.
Project description
whoop-mcp
Your WHOOP shows you charts. It can't answer "should I train hard today" or "what actually hurts my recovery". This server can.
It connects your WHOOP data to Claude, ChatGPT, and any MCP client. You ask in plain English. It computes the answer from your own records and lets the AI explain it.
You: should I train hard today?
AI: Recovery 81% (green). HRV trending up. Sleep debt cleared.
Your 7-day load is 0.94x your 28-day base.
You are conditioned for more.
Runs on your machine. Read-only. The only network peer is api.prod.whoop.com. No telemetry. MIT.
Try it in 30 seconds
You don't need a WHOOP account to see it work. Demo mode serves 150 days of realistic generated data through the exact pipeline real data uses. One line, nothing to install:
claude mcp add whoop-demo -- uvx whoop-mcp-server serve --demo
Ask "how did I sleep last week?" and watch. The patterns in the demo data are real: hard training days dent the next morning's recovery, and the correlation tool finds it.
Connect your WHOOP
Two commands. About 3 minutes, most of it WHOOP's consent screen.
uv tool install whoop-mcp-server # puts the whoop-mcp command on your PATH
whoop-mcp setup
(No uv? pipx install whoop-mcp-server works too.)
The wizard walks you through WHOOP's free developer app (it opens the dashboard and tells you exactly what to click), runs the OAuth in your browser, proves the connection with a live API call, then configures Claude Desktop, Cursor, Windsurf, VS Code, and Claude Code for you. Existing configs are backed up before any edit.
Added the server but skipped auth? Tell your AI "connect my WHOOP account". It opens the consent page from chat.
What you can ask
23 tools cover every read endpoint in WHOOP's v2 API, plus the analysis layer that turns records into answers.
| Ask | Tool behind it |
|---|---|
| "Give me the full picture of my health" | get_health_overview: status, trends, training load, records, and correlations in one call |
| "How am I doing today?" | get_daily_summary: recovery, sleep, strain, workouts for any day |
| "What actually affects my recovery?" | get_correlations: strain vs next-morning recovery, sleep vs recovery, with plain readings |
| "Am I overtraining?" | get_strain_trends: acute vs chronic load, per-sport breakdown |
| "Is my HRV improving?" | get_recovery_trends: direction, confidence, unusual days |
| "Show my overnight heart rate curve" | get_sleep_stream: minute-level HR and skin temp, lowest point and when |
| "This month vs last month?" | compare_periods: every metric, improved or declined |
| "My records this year?" | get_personal_records: bests, worsts, green streaks |
| "Export everything" | export_data: full history to local JSON and CSV |
| Week grids, raw records, profile | get_weekly_report, get_sleeps/workouts/cycles/recoveries, by-id tools with include_raw, get_profile |
| Connection trouble | get_connection_status, connect_whoop_account |
Dates are plain English everywhere: yesterday, last 30 days, this week, 2 years ago, 2026-05.
ChatGPT's required search and fetch tools are implemented too, plus 4 resources and 4 ready-made prompts (morning_readiness, weekly_review, sleep_coach, training_planner).
More questions worth asking: docs/PROMPTS.md.
Works with
ChatGPT setup
ChatGPT connects to remote MCP servers, so expose the HTTP transport and tunnel it:
whoop-mcp serve --transport http --port 8000 # endpoint: /mcp
ngrok http 8000
Then in ChatGPT: Settings, Apps & Connectors, Advanced settings, enable Developer mode, create a connector with https://<your-tunnel>/mcp and no auth.
Be careful here: a no-auth tunnel means anyone with the URL can read your health data. Keep the URL private and the tunnel short-lived. Claude's stdio setup never exposes anything.
Why the answers hold up
The model is maybe 20% of this product. The other 80% is making the numbers right. Models are bad at arithmetic over 90 days of records, so this server computes first and lets the model explain.
- Trend lines are fit to date-sorted series. WHOOP returns records newest-first. Fit arrival order instead and every trend reads backwards.
- Direction respects what the metric means. Rising HRV is improvement. Rising resting heart rate is not.
- Records land on calendar days using their own timezone offset. A sleep belongs to the morning you woke up, even when you travel.
- WHOOP rotates both OAuth tokens on every refresh. Refreshes here are serialized, saved before use, and shared across concurrent requests. Re-running auth rescues a live server without a restart.
- The client honors WHOOP's rate-limit headers, retries with backoff, and caps pagination. When a result is truncated or approximate, the output says so instead of pretending it's complete.
What it can't do: WHOOP's public API has no endpoints yet for Peak features like the stress monitor and healthspan. When WHOOP ships them, they land here. The overnight sensor stream isn't enabled for every account, and the tool reports that instead of failing.
102 tests, all offline against a faked WHOOP API. CI on Python 3.10 to 3.13.
CLI
whoop-mcp setup Guided setup: app, authorize, auto-configure clients
whoop-mcp serve Run the server (--demo, --transport stdio|http|sse, --host, --port)
whoop-mcp status Config and token state
whoop-mcp doctor Diagnose setup and connectivity
whoop-mcp auth Scriptable OAuth flow
whoop-mcp logout Delete tokens (--revoke also revokes at WHOOP)
Configuration
Nothing required after whoop-mcp setup. Overrides, in priority order: process env, ./.env, ~/.whoop-mcp/.env, ~/.whoop-mcp/config.json.
| Variable | Default | Purpose |
|---|---|---|
WHOOP_CLIENT_ID / WHOOP_CLIENT_SECRET |
none | WHOOP app credentials |
WHOOP_REDIRECT_URI |
http://localhost:8765/callback |
Must exactly match the dashboard |
WHOOP_MCP_DEMO |
off | 1 serves demo data (same as --demo) |
WHOOP_MCP_DIR |
~/.whoop-mcp |
Tokens, config, exports |
WHOOP_MCP_TZ |
system zone | IANA timezone for "today" and week bounds |
WHOOP_MCP_CACHE_TTL / WHOOP_MCP_TIMEOUT |
60 / 30 |
Seconds |
WHOOP_MCP_LOG_LEVEL |
INFO |
Logs go to stderr. stdout belongs to MCP |
WHOOP_ACCESS_TOKEN |
none | Static token for testing, no refresh |
Privacy
Read-only against WHOOP. Tokens stored locally with 0600 permissions. The only writes are local files you ask for (export_data) and the OAuth flow you trigger. One thing worth knowing: aggregates like trends and correlations can reveal more about you than single records. Connect this only to AI clients you trust with health data. Full policy: PRIVACY.md. Vulnerabilities: SECURITY.md.
The WHOOP developer dashboard asks for a privacy policy URL when you create your app. Use this repo's policy: https://github.com/rajdeepmondaldotcom/whoop-mcp/blob/main/PRIVACY.md (or your own).
Contributing
uv venv && uv pip install -e ".[dev]" && pytest && ruff check .
The whole test suite runs offline. CONTRIBUTING.md lists the invariants that keep the data honest. The short version of the architecture: client.py talks to WHOOP, transform.py cleans the records, summaries.py and analytics.py do the math, server.py exposes the tools. oauth.py and tokens.py own auth. demo.py is the generated account.
Troubleshooting
| Symptom | Fix |
|---|---|
| "WHOOP authorization required" | whoop-mcp setup, or ask your AI to connect your WHOOP account |
| Redirect fails after consent | The dashboard redirect URI must be exactly http://localhost:8765/callback |
403 missing scope |
Enable all read scopes plus offline on the app, then re-run whoop-mcp auth |
| Tools missing in a client | Re-run whoop-mcp setup, then fully restart the client |
| No recovery shown today | WHOOP scores it after you wake and sync. The summary says so |
| Sleep stream "not available" | WHOOP doesn't expose it for every account. Nightly summaries still work |
| Anything else | whoop-mcp doctor, or get_connection_status from chat, or open an issue |
License
MIT. See LICENSE. Not affiliated with or endorsed by WHOOP. WHOOP is a trademark of WHOOP, Inc.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file whoop_mcp_server-0.0.2.tar.gz.
File metadata
- Download URL: whoop_mcp_server-0.0.2.tar.gz
- Upload date:
- Size: 147.5 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
b8f9705b0a55078a0a17ec16a855bada298f8042481ddfc123e53958288e1970
|
|
| MD5 |
79eac6793652ed9dbb60c94385c0901f
|
|
| BLAKE2b-256 |
6fbce225585f9541c3b4a2ea360ec4513841a425ce7d5d732d697a66a01af826
|
File details
Details for the file whoop_mcp_server-0.0.2-py3-none-any.whl.
File metadata
- Download URL: whoop_mcp_server-0.0.2-py3-none-any.whl
- Upload date:
- Size: 67.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
50b1c39f8f567f86e795ac8b7a249fe6be9335e7bba2cf5b35587a32a7440d5e
|
|
| MD5 |
2f15827957522b99082dd27da24a7f37
|
|
| BLAKE2b-256 |
ffee84a5d303fe0cb7559545a5fa30ec477f99dc2b2cfa323b75071273b7fabf
|