Miro MCP

Miro MCP connects an AI assistant to Miro boards through the Model Context Protocol, so a whiteboard can be read and built programmatically from a chat client or coding tool rather than dragged together by hand. The server exposes boards and the items on them. An agent can search and list boards, create a new board, and read the contents of an existing one. On a board it can create, update and delete sticky notes, shapes, text, cards, frames, images and connectors, set position and size, and nest items inside frames through parent and child relationships. Layout tools operate on many items at once, so a generated structure lands in a readable arrangement instead of a pile at the origin. Code widgets place syntax highlighted source on the canvas, diagram generation turns structured input into a drawn diagram, and comments left on items can be read back. Typical uses run in both directions. Outbound, an assistant turns a document, a codebase map or a set of requirements into a diagram, a user flow or a retrospective board. Inbound, it reads a workshop board full of sticky notes and comments and summarizes or clusters them back into text. The tool surface is large, so most deployments select a reduced profile of the most used tools to keep the agent's context budget under control.