This project provides an MCP (Model Context Protocol) server and toolkit for Flutter and Dart VM applications, enabling AI coding assistants to integrate deeply with Flutter development workflows. It targets Flutter developers seeking to leverage AI for enhanced productivity, offering dynamic tooling and richer context for AI agents like Cursor, Claude, and Cline.

How It Works

The core innovation is a Flutter-native MCP server that supports dynamic tool registration. Unlike the official Dart MCP server, this project allows Flutter applications to register custom tools and resources at runtime via the mcp_toolkit package. This approach leverages the Dart VM Service Extensions to expose app state, errors, and custom functionalities directly to AI assistants, facilitating more intelligent code analysis and generation.

Quick Start & Requirements

Getting started is experimental. Users are directed to configure the MCP server using AI agents with a specific prompt and link to llm_install.md. Detailed setup and configuration guides are available in QUICK_START.md and CONFIGURATION.md. Key requirements include running the Flutter app in debug mode and a compatible Dart SDK version (e.g., 3.10.0). A Dockerfile is provided but untested.

Highlighted Details

• Dynamic Tools Registration: Enables Flutter apps to register custom tools and resources at runtime, enhancing AI assistant capabilities.
• Core AI Agent Tools: Includes get_app_errors for precise, filtered error retrieval and view_screenshot for capturing application visuals. get_view_details provides screen metrics.
• VM Service Auto-Reconnect: Automatically re-establishes connection when the Flutter app restarts.
• Smithery Registry: Registered for discoverability by AI tools via a standardized interface.
• MseeP.ai Verified: The MCP server has been verified by MseeP.ai.

Maintenance & Community

The project lists several contributors and encourages community contributions through GitHub issues and pull requests. Specific community channels like Discord or Slack are not explicitly mentioned in the README.

Licensing & Compatibility

The project is released under the MIT License, permitting free use, modification, and distribution, including for commercial purposes.

Limitations & Caveats

The "Getting Started" section is marked as experimental. The get_app_errors tool has limited platform testing (macOS, iOS only; Android, Windows, Linux, Web untested). Dump RPC methods are disabled by default due to potential token overload and require explicit enabling. The provided Dockerfile is untested.