turnnoblindeye/wellness-project-mcp

**Bring connected health and fitness data into Claude, ChatGPT, Gemini Spark, Grok, Mistral, and other MCP clients.** Wellness Project is a hosted Model Context Protocol server. Your devices and logs sync into one account, and supported assistants can read or update workouts, nutrition, sleep, recovery, body metrics, goals, labs, wellbeing, and more in plain English.
This repository is the public documentation and catalog for the hosted server. It is not a self-hosted backend and contains no private server implementation.
- App: https://wellnessproject.ai
- MCP endpoint: https://wellnessproject.ai/api/mcp
- **73 advertised public tools** generated from the production tool registry
- **15 active interactive widgets** for compatible MCP Apps UI clients
- OAuth 2.1 authentication with Dynamic Client Registration
---
What you can ask
Once connected, ask naturally and the assistant can call the matching Wellness Project tool:
- "How did my training look this week?"
- "Show my sleep for the last month."
- "How is my bench progressing?"
- "Log today's lunch: chicken burrito bowl, about 700 calories."
- "What was my resting heart rate trend vs my HRV this quarter?"
- "How am I doing against my goals?"
- "Log 16 ounces of water."
Read tools return your logged and synced health data. Write tools add, update, or delete entries when you ask. Widget tools can render structured interactive views instead of a wall of numbers.
Connect Wellness Project to an MCP client
First create an account at https://wellnessproject.ai and connect the health sources you want under **Settings → Devices**. Every client below uses the same hosted endpoint:
https://wellnessproject.ai/api/mcp
Authentication is handled by OAuth. Personal API keys are not part of the current public connection flow.
Claude
1. Open Claude on the web.
2. Go to **Customize → Connectors → Add custom connector**.
3. Paste https://wellnessproject.ai/api/mcp.
4. Leave Client ID and Client Secret blank if Claude shows those fields. Claude can register itself dynamically.
5. Sign in to Wellness Project and authorize the connection.
Claude Code
Claude Code installs Wellness Project as a plugin. The plugin carries nothing but
this server's configuration and a setup skill.
Restart Claude Code, then run /mcp, select **wellness-project**, and authorize.
There is no client ID, client secret, or API key to enter. Run
/wellness-project:setup if you would rather be walked through it.
Cline
Cline connects to Wellness Project as a remote MCP server. Nothing is cloned,
installed, or run locally.
1. Open the **MCP Servers** icon in the Cline sidebar and choose **Remote Servers**.
2. Enter a server name, for example wellness-project.
3. Paste https://wellnessproject.ai/api/mcp as the server URL and add it.
4. Complete the Wellness Project OAuth flow when Cline opens it in your browser.
To edit cline_mcp_settings.json directly instead:
Set "type": "streamableHttp" explicitly. Cline falls back to the legacy SSE
transport when the type is omitted. There is no client ID, client secret, or API
key to enter. llms-install.md has the same steps written for
an agent doing the setup.
ChatGPT
ChatGPT custom MCP apps use developer mode. Current OpenAI availability is plan-dependent: full MCP support including write/modify actions is available in beta for Business, Enterprise, and Edu, while Pro can connect MCPs with read/fetch permissions in developer mode.
1. Enable developer mode for your account or workspace as permitted by your ChatGPT plan.
2. Create a custom MCP app from ChatGPT's Apps settings.
3. Use https://wellnessproject.ai/api/mcp as the remote MCP server.
4. Complete the Wellness Project OAuth flow.
OpenAI's current setup and plan details: https://help.openai.com/en/articles/12584461