A2A MCP News
mcp-server1

Dart and Flutter MCP Server

The official Dart and Flutter MCP Server from the Flutter team. It connects AI coding assistants like Claude and Cursor to your Dart/Flutter projects, enabling natural language commands for code analysis, running Flutter CLI tools, debugging, and interacting with running apps.

Overview

The Dart and Flutter MCP Server is the official implementation of the Model Context Protocol (MCP) developed and maintained by the Dart and Flutter teams at Google. It allows AI coding assistants (such as Claude, Cursor, Gemini CLI, and others) to interact deeply with your Dart and Flutter development environment.

Requiring Dart SDK 3.9+ or Flutter 3.35+, this server exposes powerful tools that give AI direct access to project context, CLI commands, analyzer results, and even live running applications — making it one of the most advanced official MCP integrations available.

Key Features

  • Project & Code Analysis: Deep insights into your codebase, dependencies, widget tree, and structure.
  • Flutter CLI Integration: Run commands like flutter pub get, flutter analyze, flutter test, flutter format, and more.
  • Live App Interaction: Connect to running Flutter apps for screenshots, widget inspection, hot reload/restart, and state debugging.
  • DevTools-like Capabilities: Access tooling normally available only in IDEs or Flutter DevTools.
  • Experimental but Actively Developed: Continuously improved based on community feedback.
  • Stdio Transport Support: Works with all major MCP clients.

Use Cases

  • AI-Assisted Development: Ask your AI to "fix this layout bug", "add dark mode support", or "optimize this widget" with full project context.
  • Debugging & Troubleshooting: "Why is this screen crashing on Android?" or "Capture a screenshot of the current UI state".
  • Rapid Feature Building: Generate new screens, refactor code, or run tests conversationally.
  • Learning Flutter: Get guided help while building apps, even as a beginner.
  • End-to-End Workflows: Combine with other MCP servers for complete AI-orchestrated mobile development.

Installation & Setup

Prerequisites

  • Dart SDK 3.9+ or Flutter 3.35 (beta or stable channel recommended).

Quick Setup with Cursor or Claude

Most modern AI clients support easy addition. For Cursor, many users simply use the built-in "Add MCP Server" flow pointing to the official repo.

For Claude Code or manual configuration, use stdio transport:

claude mcp add dart -- dart mcp-server

Or add to your client config (example for Cursor/Claude Desktop):

{
  "mcpServers": {
    "dart-flutter": {
      "command": "dart",
      "args": ["mcp-server"]
    }
  }
}

Detailed setup instructions are available in the official Flutter documentation and the repository README.

Popular Community Complements

While the official server is the recommended starting point, the community has built excellent extensions:

  • mcp_flutter (Arenukvern) — Dynamic tool registration from running Flutter apps.
  • flutter-mcp (adamsmaka) — Real-time Flutter/Dart docs and pub.dev package search.
  • Marionette MCP — Runtime UI interaction (taps, scrolls, screenshots) for testing.

Compatibility

  • Clients: Cursor, Claude Desktop/Code, Gemini CLI, GitHub Copilot, and any stdio-compatible MCP client.
  • License: Open-source (part of the official Dart/Flutter ecosystem).

This official MCP server significantly lowers the barrier for AI-augmented Flutter development and represents the Flutter team's strong commitment to agentic AI workflows in 2026.

Tags

flutterdartmcpofficialai-integrationclaudecursordevtoolsflutter-development