A powerful Model Context Protocol (MCP) server that enables AI agents to control Android devices via ADB with intelligent self-learning, UI pattern memory, vision-enhanced capabilities, and macro support.
- Intelligent Self-Learning & UI Pattern Memory (SQLite Knowledge Base): AI learns and remembers thousands of actions and UI patterns, making control smarter and more accurate over time.
- Auto-Recovery: Automatically recalculates button coordinates if UI elements move, ensuring actions are always precise even with minor UI changes.
- Vision-Enhanced (Image Recognition): Powered by
sharp, the AI can find and interact with UI elements based on image templates, perfect for Games or apps without clear XML structures. - Macro Support: Execute complex sequences of actions automatically and smoothly, becoming more optimized and flexible with self-learning.
- Semantic UI Mapping: Automatically recognizes buttons via
resource-idortext, ensuring the AI doesn't get lost when apps update their interface. - Batch Execution & Persistent Connection: Optimized ADB connection for near-instant command response and minimized local network latency.
- Requirements: Node.js v18+, ADB installed and available in your PATH.
- Install Dependencies:
npm install
- Build:
npx tsc
- Start:
node dist/index.js
adb_devices: List connected devices.ui_learn_semantic: Teach the AI to recognize buttons intelligently via ID/Text.ui_save_template: Save an image file as a template for vision tasks.adb_find_image: Find coordinates of a template image on the current screen.adb_tap_image: Find and tap an image template in one command.adb_smart_action: Execute actions using learned knowledge (auto-finds buttons).adb_batch_shell: Run optimized shell command sequences.adb_dump_ui: Explore the UI hierarchy of unknown applications.
We welcome contributions! Please see our CONTRIBUTING.md for guidelines on how to get started, report bugs, and suggest new features.
If you discover any security vulnerabilities, please report them responsibly. Refer to our SECURITY.md for details on how to report and our supported versions.
This project is licensed under the MIT License. See the LICENSE file for more details.
The project includes a built-in GitHub Action (.github/workflows/release.yml) to automatically package and create a GitHub Release whenever you push a version tag (e.g., v1.0.0).
Rất cảm ơn các anh em đã dành thời gian đóng góp cho dự án này! Mọi sự giúp đỡ của anh em đều vô cùng quý giá. ✨
| Contributor | Role | Contribution |
|---|---|---|
| tiendung2k03 | Project Lead | Original Idea & Development |
| Huynh Nhu | Supporter | Documentation & Community |
Anh em muốn có tên trong bảng vàng này thì đừng ngần ngại gửi Pull Request nhé! 🎀🧸
Anh em có bất kỳ thắc mắc, góp ý hay muốn hợp tác thì cứ liên hệ mình qua các kênh này nhé:
- Email: monstertattoovn696@gmail.com 📧
- Facebook: Tiến Dũng 👤
- Zalo: 0766459014 💬
- Telegram: @tiendung2k03
✈️
Rất mong được giao lưu cùng anh em! 🙏✨