mcpField
M
Miro

Miro

Remote
github.com/miroappMedia & Design164Updated 6w ago

About Miro

Miro AI Developer Tools connects AI coding assistants to Miro boards through an MCP server, plugins and agent skills. It targets developers who want to create diagrams, extract design context or track tasks through conversation.

Key features

  1. 01Access and edit Miro boards through an MCP server
  2. 02Generate flowcharts, sequence diagrams and ERDs
  3. 03Convert board designs and wireframes into code
  4. 04Sync tasks between AI chats and Miro tables
  5. 05Plugins for Claude Code, Gemini CLI and Codex

Use cases

  1. 01Generating code from wireframes on a board
  2. 02Creating architecture diagrams of a codebase on a board
  3. 03Running visual code reviews of PRs or branch changes

MCP support

The project provides a hosted Miro MCP server that AI clients connect to after OAuth authorization, and includes MCP configuration instructions for clients such as Cursor and VSCode.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.miro.com/
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

Miro AI Developer Tools

Documentation

Connect AI coding assistants to your Miro boards. Create diagrams, extract context, generate code from designs, and track tasks—all through natural conversation.

Gemini CLI docs, diagram, table demo


Share your feedback.


What's in This Repository?

This repo provides everything you need to connect AI tools to Miro:

ComponentWhat It Does
Miro MCP ServerAPI that gives AI agents access to your Miro boards
Plugins & ExtensionsPre-built integrations for popular AI tools
DocumentationGuides for using and developing integrations
Supported AI Tools
AI ToolIntegration
Claude CodePlugins
Gemini CLIExtensions
CodexLocal plugins
KiroPower
Agent SkillsSkills
Cursor, VSCode, Windsurf, etc.MCP Config

Quick Start

Step 1: Choose Your AI Tool

Select your AI tool below and follow the installation steps.

Step 2: Install
Claude Code (Recommended)
/plugin marketplace add miroapp/miro-ai
/plugin install miro@miro-ai

The miro plugin bundles all skills (browse, code-review, code-spec, code-explain-on-board, diagram, doc, format, table) and the Miro MCP server.

Restart Claude Code after installation. If you previously configured Miro MCP manually, remove the duplicate to avoid conflicts — the plugin already manages the MCP connection for you.

See Claude Code Plugins for full documentation.

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

Gemini CLI installs the repo as an extension, registers the Miro MCP server from gemini-extension.json, and auto-discovers the 8 bundled skills from skills/. Restart Gemini CLI and authenticate when prompted.

See Gemini CLI Extension | Official Docs

Kiro
  1. Open the Powers panel in Kiro
  2. Click Add power from GitHub
  3. Enter: miroapp/miro-ai and select powers/code-gen

For local development, see CONTRIBUTING.md.

See Kiro Powers | Official Docs

Agent Skills (agentskills.io)

Portable knowledge skills that work across AI coding tools (Claude Code, Cursor, VS Code + Copilot, Codex, etc.).

npx skills add miroapp/miro-ai                       # Interactive install
npx skills add miroapp/miro-ai --list                # List available skills
npx skills add miroapp/miro-ai --skill=miro-code-review   # Install specific skill

Available skills: miro-code-explain-on-board, miro-code-review, miro-code-spec

See Agent Skills Overview | agentskills.io

Other MCP Clients (Cursor, VSCode + Copilot, Windsurf, Lovable, Replit, etc.)

Add to your MCP client configuration file:

{
  "mcpServers": {
    "miro": {
      "url": "https://mcp.miro.com/"
    }
  }
}

Note: If a Miro plugin or extension becomes available for your client later, switch to it and remove the manual entry to avoid duplicate servers.

See MCP Setup Guide for client-specific paths.

Step 3: Authenticate

When you first use a Miro command, you'll be prompted to authenticate:

  1. A browser window opens with Miro OAuth
  2. Log in to your Miro account
  3. Grant access to the requested boards
  4. Return to your AI tool—you're connected!
Step 4: Try It

Test your setup with these example prompts:

"What's on my Miro board https://miro.com/app/board/..."
"Create a flowchart on Miro showing a user login flow"
"Summarize the architecture diagram on my Miro board"
Developer Mode

Want to modify plugins, test changes locally, or build your own? See CONTRIBUTING.md for dev setup instructions.

Codex plugins are generated locally into codex-plugins/*/ with the repo marketplace at .agents/plugins/marketplace.json. See Codex Plugins for the generated structure and CONTRIBUTING.md for regeneration and local testing.

Avoiding Duplicate Servers

If your client has a Miro plugin or extension, use only that — do not also add https://mcp.miro.com/ manually. Each installation method creates an independent MCP connection with its own OAuth session, and running more than one causes duplicate tools and auth confusion.

See the Duplicate MCP Servers guide in Miro's Developer docs for diagnosis and cleanup steps.


Enterprise Users

Admin Approval Required: If your organization is on a Miro Enterprise plan, your admin must enable the MCP Server before you can connect.

See Enterprise Guide for setup instructions.


What Can You Do?

Skills (Claude Code)

Skills auto-activate when you describe what you want naturally. No slash commands required — just include a Miro board URL.

SkillTriggers On
miro-code-explain-on-board"Explain this codebase on …", "Visualize the architecture on …"
miro-code-review"Review PR 123 on …", "Visual review of this branch on …"
miro-code-spec"Extract specs from …", "Pull the design docs at …"

Creating content on a board — diagrams, documents, tables, stickies, standalone formats — needs no skill. Ask for it directly; the Miro MCP server's own tools cover it and load their authoring guidance on demand.

Capabilities (All Platforms)
CapabilityDescription
Generate DiagramsCreate flowcharts, sequence diagrams, ERDs from code or descriptions
Generate CodeConvert board designs and wireframes into working code
Track TasksSync tasks between AI conversations and Miro tables
Extract ContextRead board content to inform AI workflows

Available Plugins

Claude Code
PluginDescription
miroCore MCP integration with 3 bundled skills (code-explain-on-board, code-review, code-spec)
Gemini CLI
ExtensionDescription
miroCore MCP integration with bundled skills
Codex
PluginDescription
miroCore MCP integration with native $miro:<skill> activation and MCP access
Agent Skills
SkillDescription
miro-code-explain-on-boardExplain a codebase on a board as a minimal, notation-correct diagram set plus a companion doc
miro-code-reviewVisual code review from PRs, local changes, or branch comparisons
miro-code-specExtract Miro specs (docs, diagrams, prototypes, tables, images) to .miro/specs/
Kiro
PowerDescription
code-genDesign-to-code workflows

Security

  • OAuth 2.1 secure authorization
  • Enterprise compliance standards
  • Permission-based access via your Miro account
  • Rate limited API protection

Documentation

Getting Started
Platform Guides
Reference
Development
External

Supported MCP Clients

Claude Code, Cursor, VSCode + GitHub Copilot, Gemini CLI, Lovable, Replit, Windsurf, Kiro, Glean, Devin, OpenAI Codex


Feedback & Support


License

MIT — see LICENSE


Built with love by Miro

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.