Use the Wheel of Life inside your AI assistant.
LifeWheel runs as a callable tool, not just a webpage. Any AI assistant that speaks the Model Context Protocol (MCP) — ChatGPT, Claude Desktop, Cursor, Continue, Cline, Goose — can connect directly to LifeWheel and let users do a Wheel of Life check-in, create habits, log moods, and journal without leaving the chat. Three demo tools work without any signup. The full toolset unlocks with one OAuth click.
What this is for
For end users: if you live in ChatGPT Plus or Claude all day and want a quick check-in on your life balance without installing another app, connect LifeWheel and just ask. The AI assistant becomes the interface; the wheel and habits live in your account.
For coaches: run a Wheel of Life intake inside a ChatGPT Plus session with a client. The client doesn't need a LifeWheel account — the demo tools render a full wheel from scores you discuss in the room. For ongoing work, the client signs in via OAuth and you can guide future sessions using their actual data.
For AI builders: if you're building a wellness assistant, intake bot, or personal-development workflow, the LifeWheel MCP gives you a tested 21-tool surface for life-balance assessment, habit creation, journaling, mood tracking, and decision support. Per-tool annotations (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) make agent reasoning predictable.
Compatible AI clients
ChatGPT (Apps SDK)
Add as a connector in ChatGPT. Works in Plus and Team plans. Discoverable in the apps directory once approved by OpenAI.
Claude Desktop
Add a custom MCP server in Claude Desktop's settings. Use the connector URL above.
Cursor
Add LifeWheel as an MCP server in Cursor's settings. Run wheel and habit operations from inside your IDE.
Any MCP-compatible client
Continue, Cline, Goose, Zed, custom agents — anything that speaks JSON-RPC 2.0 MCP.
Quick start: try it without signup
The fastest way to see what this looks like is to use one of the three demo tools. They work without authentication and never touch any user data.
// In any MCP-connected AI client, ask:
"Use LifeWheel to start a wheel-of-life demo for me."
// The assistant calls:
start_wheel_demo({ locale: "en" })
// You see an 8-sphere quiz widget. You drag the sliders, hit "See my wheel."
// The assistant calls:
render_demo_wheel({ scores: [/* 8 sphere scores */] })
// You get a beautiful, color-per-sphere Wheel of Life rendering.
// Nothing was saved. Nothing was ever sent to lifewheel.us about you.You can also call explain_sphere({ sphere: "career", locale: "en" }) to get a short, well-written explanation of any of the 8 standard spheres — useful when introducing the methodology to a new client mid-conversation.
The full tool surface (auth required)
Once a user signs in via OAuth (one click in their AI client), the assistant can call any of these tools on their behalf. Every tool is annotated so the AI knows exactly what each call does.
| Tool | What it does | Hint |
|---|---|---|
show_wheel | Renders the user's life wheel as a widget | read |
analyze_wheel | Returns weakest/strongest spheres + recent deltas | read |
weekly_review | Last-7-day summary: tasks done, journal, mood, wheel changes | read |
list_habits | Active habits with today's completion state | read |
search | Free-text search across goals, journal, mood, decisions | read |
fetch / show_goal | Get a single record by id | read |
whoami | Account diagnostic: uid, sphere/goal/wheel counts | read |
create_goal / create_habit | Create a new goal or recurring habit with a sphere link | append |
create_todo / create_journal | Add a task or journal entry | append |
log_mood / create_vision_item | Log a mood or add a Vision Space tile | append |
complete_goal / toggle_todo | Mark a goal achieved or a task complete | set state (idempotent) |
pause_goal / update_goal / update_vision_item | Pause, resume, or edit existing records | set state (idempotent) |
set_wheel_slider | Set a sphere's current value with history preserved | set state |
commit_decision | Mark a Decide Mode decision as committed | set state |
delete_goal | Permanently remove a goal (only destructive operation) | destructive |
Per-tool documentation including parameter schemas is published in the MCP server's tools/list response — any connected agent will discover them automatically on connection.
How auth works
OAuth 2.1 with PKCE. When a user adds LifeWheel as a connector in their AI client, the client opens our authorization page (Firebase-backed sign-in). On consent, we issue an access token (30-day lifetime) and refresh token (180-day). Tokens are revoked instantly when the user disconnects the connector or deletes their account.
The OAuth metadata endpoint is at https://lifewheel.us/.well-known/oauth-authorization-server — standards-compliant, so most MCP clients auto-configure.
Privacy
The integration is opt-in and inactive until the user explicitly adds LifeWheel as a connector. Connector data is never used to train AI models or for advertising. Tool inputs and outputs are not stored on our servers beyond the in-flight HTTP request. We maintain a per-user audit log of write operations so users can review what an AI did under their account. Full disclosure: see our privacy policy — specifically the "ChatGPT Integration (OpenAI Apps SDK)" section.
Why the unauthenticated demo matters. Most apps require signup before letting an AI assistant do anything useful. The three LifeWheel demo tools were built so a person can experience the entire core methodology — rate 8 spheres, see the wheel, get an explanation — inside ChatGPT or Claude before deciding whether they want to install our iOS app. That low-friction discovery is the whole point.
FAQ
What AI assistants can I use LifeWheel with?
https://lifewheel.us/mcp.Do I need a LifeWheel account to use the MCP integration?
start_wheel_demo, render_demo_wheel, explain_sphere) work without any signup or authentication. Anyone with an MCP-compatible client can run a complete Wheel of Life session inside their AI assistant without creating an account. Sign-in via OAuth 2.1 unlocks reading and writing your real data.Is the MCP integration free?
What can the AI do once I'm signed in?
delete_goal, which requires explicit confirmation. Every tool is annotated with explicit readOnly, destructive, idempotent, and openWorld hints so the AI client knows exactly what each call does.How is privacy handled?
Can coaches use this with clients?
Try it
Add the connector URL https://lifewheel.us/mcp in your AI client, or just download the iOS app to get the wheel on your phone.