Embedded MCP Overview

Connect a server-side AI agent to MindCloud apps for one embedded end user. Explore setup, tool calls, and security.

MindCloud Embedded MCP lets your server-side AI agent use an end user's connected apps. The agent can find apps and actions, inspect inputs, run actions, and help that end user connect an account.

Each MCP URL belongs to one MindCloud end user. Your server selects the end user from its signed-in session. Keep the MindCloud API key and MCP client on your server.

The server is in beta. It uses stateless Streamable HTTP at https://connect.mindcloud.co/v2/embedded/end-users/{endUserId}/mcp.

Start here

Set up Embedded MCP to enable the server, create an end user, and connect your agent.
Understand tool calls to learn the response format and choose a tool.
Open a tool page for its exact inputs, output example, and next step.

Task	Tool
Find an app	`search-apps`
Find an action	`search-actions`
Read action inputs	`get-action-schema`
See connected accounts	`list-connections`
Run an action	`execute-actions`
Read a stored result	`query-result-set`
Resolve a name to an ID	`resolve-lookup`
Connect or repair an account	`get-connect-url`

The Embedded MCP example app shows a server-side agent with chat, live tools, and setup steps. Use the tool pages as the API contract.