Arenukvern/mcp_flutter

What it solves

flutter-mcp-toolkit provides a way for AI agents (such as Claude Code, Cursor, and Zed) to interact with and control a running Flutter application in debug mode. It eliminates the need for agents to guess the app's state or structure by providing a direct, programmable interface to the app's internals.

How it works

The project consists of a Dart MCP server and a Flutter package (mcp_toolkit). When added to a Flutter app, it allows an AI agent to connect via the Model Context Protocol (MCP). The agent can then use a set of predefined tools to inspect the app, take semantic snapshots, and perform user-like actions. Additionally, the toolkit allows developers to register custom MCP tools and resources directly within the Flutter app at runtime, creating a closed feedback loop between the AI and the application.

Who it’s for

Flutter developers who want to use AI agents to help with testing, debugging, and driving the application during development.

Highlights

• App Control: Ability to tap widgets, type into forms, scroll, and navigate within the running app.
• App Inspection: Take semantic snapshots, view details, read logs, and evaluate Dart expressions.
• Lifecycle Management: Support for hot-reload and hot-restart directly from the AI agent.
• Dynamic Tooling: Allows apps to expose their own custom MCP tools and resources to the AI agent at runtime.