Back to MCP Servers

Miro MCP Server

Official Miro MCP server that connects AI coding tools to Miro boards for diagram generation, code spec extraction, and visual collaboration.

Collaboration by Miro OAuth2 active
Overview

The Miro MCP Server is an official, provider-hosted Model Context Protocol endpoint that lets AI coding tools read from and write to Miro boards via natural language. It exposes around 30 tools spanning board search, item listing, diagram creation, code widget management, image handling, tables, documents, and comments, all secured through an OAuth 2.1 flow with dynamic client registration.

The server is designed around two core use cases: generating diagrams on a Miro board from PRDs, code, or free text, and generating code from existing board content such as design specs and diagrams. It supports a broad set of MCP-compatible clients including Cursor, Claude Code, VS Code with GitHub Copilot, Windsurf, Gemini CLI, Replit, Lovable, Kiro, Devin, Glean, OpenAI Codex, and Amazon Q. For Claude Code and Gemini CLI, Miro additionally ships native plugins and extensions via the miroapp/miro-ai GitHub repository, which bundle skills like miro-browse, miro-code-review, miro-code-spec, miro-diagram, miro-doc, and miro-table.

The MCP server is in public beta. The endpoint is hosted at https://mcp.miro.com/ so there is no server to self-host; clients connect directly and authenticate through Miro's OAuth screen. Enterprise Plan organizations must first have an admin enable the MCP Server for the org before users can connect.

Tools

Tool Description
board_list_items List items on a Miro board with filters and pagination.
board_search_boards Search the user's Miro boards by name or metadata.
board_create Create a new Miro board.
context_explore Explore board context for AI reasoning.
context_get Retrieve contextual board content (uses AI credits).
diagram_create Create a diagram (flowchart, sequence, ERD, UML, etc.) on a board.
diagram_get_dsl Return the DSL specification for diagrams.
doc_get Read a markdown document item from a board.
doc_create Create a markdown document on a board.
doc_update Update an existing board document.
table_list_rows List rows of a board table.
table_create Create a structured table on a board.
table_sync_rows Sync rows in an existing table.
code_widget_get Read a code widget's contents.
code_widget_create Create a code widget on a board.
code_widget_update Update a code widget.
code_widget_delete Delete a code widget.
image_get_data Fetch raw image data from a board item.
image_get_url Get the URL of a board image.
image_get_upload_url Get a signed upload URL for adding new images.
image_create Create an image item on a board.
layout_read Read a layout definition from a board.
layout_create Create a layout on a board.
layout_update Update an existing layout.
layout_get_dsl Return the DSL specification for layouts.
comment_list_comments List comments on a board or item.
comment_reply Reply to a comment thread.
comment_resolve Resolve a comment thread.
Setup Guide

Prerequisites

  • Active Miro account with access to a workspace
  • MCP client that supports OAuth 2.1 (Cursor, Claude Code, VS Code + Copilot, Windsurf, Gemini CLI, etc.)
  • Enterprise Plan: admin must enable the Miro MCP Server for the org first

Generic MCP client configuration

Add the following to your MCP client config (Claude Desktop, Cursor, Windsurf, VS Code, etc.):

{
  "mcpServers": {
    "miro-mcp": {
      "url": "https://mcp.miro.com/",
      "disabled": false,
      "autoApprove": []
    }
  }
}

On first use, a browser window opens for the Miro OAuth flow. Select the team to install Miro MCP into, approve board access, and return to your AI tool.

Claude Code (recommended)

/plugin marketplace add miroapp/miro-ai
/plugin install miro@miro-ai

Restart Claude Code after installation.

Gemini CLI

gemini extensions install https://github.com/miroapp/miro-ai

Agent Skills

npx skills add miroapp/miro-ai
Use Cases
  • Generate flowcharts, sequence diagrams, and ERDs on a Miro board directly from a codebase, PR diff, or PRD text
  • Extract design specifications and layouts from a Miro board into local files to drive code generation in Cursor or Claude Code
  • Run visual code reviews from inside an AI coding tool by syncing PRs or branch comparisons onto a board
  • Create and update markdown docs, structured tables, and code widgets on boards as the agent makes progress on a task
  • Browse and summarize the contents of a board (items, comments, layouts) so an agent has shared context with the design team
Example Prompts
  • "Read the PRD on my Miro board called 'Checkout v2' and generate a sequence diagram for the payment flow."
  • "Look at the architecture diagram on my Miro board and scaffold the matching Node.js services in this repo."
  • "List all sticky notes on the 'Sprint Planning' board and turn them into a structured table grouped by owner."
  • "Pull the design spec from the Miro board, then implement the React component and add a code widget back to the board with the result."
  • "Review the open comments on my Miro board and reply to any that ask for technical clarification."
Pros
  • Official server hosted by Miro at https://mcp.miro.com/, no self-hosting required
  • Broad coverage with around 30 tools spanning diagrams, docs, tables, images, code widgets, layouts, and comments
  • OAuth 2.1 with dynamic client registration and enterprise compliance controls
  • First-class plugins and extensions for Claude Code and Gemini CLI in the miroapp/miro-ai repo
Limitations
  • Still in public beta, so tool behavior and rate limits may change
  • Enterprise Plan organizations require admin enablement before users can connect
  • Some tools such as context_get consume Miro AI credits, which can add cost on top of an existing plan
Alternatives