Mouve MCP Connector
Bring your Mouve fitness and nutrition data into Claude, ChatGPT, or any Model Context Protocol client. Remote MCP server, OAuth 2.1, user-scoped.
What it is
The Mouve MCP connector is a remote Model Context Protocol server that exposes a curated set of Mouve tools to authorized AI agents. Each request runs in the user’s security context: an AI agent can only see and modify data the user explicitly authorized through the Mouve OAuth consent screen.
Server URL: https://api.mouve.it/mcp
How to connect
Fastest way: give your AI the Connect your AI link and it walks you through it for your app.
- Claude Desktop / Claude.ai: add Mouve as a remote connector and point it at
https://api.mouve.it/mcp. Claude will discover the authorization server, open the consent screen, and complete the OAuth 2.1 flow on your behalf. - ChatGPT Apps / GPTs: register Mouve as an MCP source with the same server URL. ChatGPT runs the Dynamic Client Registration (DCR) flow automatically.
- Custom MCP clients: implement OAuth 2.1 with PKCE against the discovery metadata described below. Register your client dynamically at
POST https://api.mouve.it/oauth/register(RFC 7591).
OAuth endpoints
- Authorization server metadata (RFC 8414):
https://api.mouve.it/.well-known/oauth-authorization-server - Protected resource metadata (RFC 9728):
https://api.mouve.it/.well-known/oauth-protected-resource/mcp - Dynamic Client Registration (RFC 7591):
POST https://api.mouve.it/oauth/register - Token revocation (RFC 7009):
POST https://api.mouve.it/oauth/revoke
Available tools
The connector ships 21 tools grouped by domain. The canonical schema is served by the MCP tools/list call once you are connected; this page is a high-level summary.
Profile
Read public profile, bio, avatar, display name.
Workouts
List history, inspect sessions, create or log workouts from an AI agent.
Nutrition
Parse food descriptions, analyze food photos, generate meal suggestions and daily meal plans, log meals.
Feed
Read the user’s own posts and saved posts.
Memory
Per-user memory notes so agents keep context across conversations.
Design system
Read the Mouve design tokens and component guidance (read-only, non-sensitive).
Utilities
Health checks, metadata helpers, and other small tools used by clients during setup.
Scopes
Every tool declares the scopes it requires. The consent screen shows the union of scopes an AI client is asking for; the user can approve or cancel. The canonical list of scopes, and the tools mapped to each one, is published in the MCP server’s resource metadata (scopes_supported). Scopes follow <domain>.<read|write> with exact-match semantics — write never implies read. Domains: profile, body, follows, posts, workouts, nutrition, notifications, and memory. Body, workout, and nutrition scopes cover health data and require explicit consent.
Rate limits
60 requests per minute per user by default. Limits are applied at the MCP server and may be tightened in response to abusive traffic. Authorization, discovery, and health endpoints are exempt.
Support and security
- Product and integration support: [email protected]
- Security reports: [email protected] (also published at /.well-known/security.txt)
Legal
- MCP Connector Privacy Policy
- MCP Connector Terms of Service
- Mouve Developer Terms — binding on developers and operators of apps, agents, and integrations that access the Mouve API
- Main Mouve Privacy Policy
- Main Mouve Terms of Service