Android Autonomous AI Agent

An intelligent, multimodal agent that "sees" an Android screen using ADB and "thinks" to autonomously perform tasks on a mobile device.

Architecture Overview

The system follows a See-Think-Act loop:

1. Observer: Captures a screenshot via ADB.
2. Brain: Sends the image + goal to an AI model
3. Executor: Translates AI decisions into physical taps and swipes.

Project Structure

• core/: The logic engine (Analyzer, Generator, Agent).
• utils/: Helper utilities for logging and APK manipulation.
• data/: Storage for APKs, screenshots, and execution logs.

Getting Started

1. Prerequisites

• Python 3.10+
• ADB (Android Debug Bridge) installed and in your PATH.
• Apktool (for the Analyzer module).

2. Setup

1. Clone the repository.
2. Install dependencies:

   pip install -r requirements.txt

3. Create a .env file in the root:

   GEMINI_API_KEY=your_api_key_here
   ANDROID_DEVICE_ID=your_device_serial