mcp_flutter
MCP Toolkit for Flutter AI Agent Driven Development (MCP/CLI + custom client side tools) - via closed feedback loop (visual & semantic snapshot) and high client
What is mcp_flutter?
MCP Toolkit for Flutter AI Agent Driven Development (MCP/CLI + custom client side tools) - via closed feedback loop (visual & semantic snapshot) and high client side customization adaptable for any Flutter app. Nowadays it is often called as agentic harness.
Overview of mcp_flutter
What it is
flutter-mcp-toolkit is a Dart MCP server and Flutter package that allows AI agents (such as Codex, Zed, Cursor, Intent, Claude Code, Cline) to inspect and interact with a running Flutter application. It provides a closed feedback loop between the AI agent and the app, enabling the agent to take semantic snapshots, tap widgets, type into forms, hot-reload, and read logs. Applications can also register their own MCP tools and resources at runtime. The toolkit is becoming a chain of declarative automation APIs, using the same observe/act/verify vocabulary across interactive MCP, the CLI, programmatic Dart scenarios (flutter_mcp_harness), and declarative build/lifecycle runners like oka, all over a universal AutomationDriver contract.
Key features
- AI agent can inspect and drive a running Flutter app in real time.
- Supports semantic snapshots, widget tapping, form typing, hot reload, and log reading.
- Applications can register custom MCP tools and resources at runtime without leaving the conversation.
- Includes a CLI (flutter-mcp-toolkit) for setup and repeated automation loops.
- Provides a universal AutomationDriver contract used across MCP, CLI, harness, and oka runners.
- Has published packages flutter_mcp_harness and mcp_toolkit on pub.dev.
- Can be installed via Docker or MCP registries (GHCR OCI image).
Use cases
- AI-assisted development for Flutter apps, allowing agents to test and debug interactively.
- Automated UI inspection and verification during development.
- Integrating custom runtime tools for AI agents without restarting the app.
- Declarative automation chains spanning multiple runners (MCP, CLI, harness, oka).
Getting started
bash # 1. Install the binary curl -fsSL https://raw.githubusercontent.com/Arenukvern/mcp_flutter/main/install.sh | bash # 2. Add the toolkit to your Flutter app cd my-flutter-app flutter-mcp-toolkit codegen-init # 3. Install skills for your AI agent flutter-mcp-toolkit init claude-code # or: cursor | codex | cline | agents-skills | all # 4. Run flutter run --debug
License and pricing
The license is MIT. The project is open source.
這是什麼
flutter-mcp-toolkit 是一個 Dart MCP 伺服器和 Flutter 套件,允許 AI 代理(如 Codex、Zed、Cursor、Intent、Claude Code、Cline)檢查並與正在執行的 Flutter 應用程式互動。它在 AI 代理和應用程式之間提供了一個封閉的回饋迴圈,代理可以擷取語義快照、點擊小部件、在表單中輸入文字、熱重載以及讀取日誌。應用程式也可以在執行時註冊自己的 MCP 工具和資源。該工具包正逐漸成為一系列宣告式自動化 API,在互動式 MCP、CLI、程式化 Dart 場景(flutter_mcp_harness)以及像 oka 這樣的宣告式建構/生命週期執行器中使用相同的觀察/行動/驗證詞彙,全部基於通用的 AutomationDriver 協定。
主要功能
- AI 代理可以即時檢查並驅動正在執行的 Flutter 應用程式。
- 支援語義快照、小部件點擊、表單輸入、熱重載和日誌讀取。
- 應用程式可以在執行時註冊自訂的 MCP 工具和資源,無需離開對話。
- 包含用於設定和重複自動化迴圈的 CLI(flutter-mcp-toolkit)。
- 提供通用的 AutomationDriver 協定,用於 MCP、CLI、harness 和 oka 執行器。
- 已發布 flutter_mcp_harness 和 mcp_toolkit 套件於 pub.dev。
- 可透過 Docker 或 MCP 註冊中心(GHCR OCI 映像檔)安裝。
使用情境
- Flutter 應用的 AI 輔助開發,允許代理互動式測試和除錯。
- 開發期間的自動化 UI 檢查與驗證。
- 無需重新啟動應用程式即可為 AI 代理整合自訂執行時期工具。
- 跨多個執行器(MCP、CLI、harness、oka)的宣告式自動化鏈。
快速開始
bash # 1. 安裝二進位檔案 curl -fsSL https://raw.githubusercontent.com/Arenukvern/mcp_flutter/main/install.sh | bash # 2. 將工具包加入你的 Flutter 應用程式 cd my-flutter-app flutter-mcp-toolkit codegen-init # 3. 為你的 AI 代理安裝技能 flutter-mcp-toolkit init claude-code # 或:cursor | codex | cline | agents-skills | all # 4. 執行 flutter run --debug
授權與價格
授權為 MIT。此專案是開放原始碼。
Based on the project's public README and repository metadata.