​
Welcome to DroidRun

DroidRun is a powerful framework that enables you to control Android devices through LLM agents. It provides a simple and intuitive way to automate Android device interactions using natural language commands.

​
Features

Natural Language Control

Control your Android device using natural language commands

Multiple LLM Support

Support for OpenAI, Anthropic, and Google Gemini

Simple CLI

Easy-to-use command line interface

Python SDK

Comprehensive SDK for custom automation tasks

​
Quick Example

# Simple CLI usage
droidrun "Open the settings app"

# With specific provider
droidrun "Open calculator app" --provider gemini --model gemini-2.0-flash

# With vision capabilities
droidrun "Open Calculator and take a screenshot" --vision

Or with Python:

import asyncio
import os
from droidrun.agent.react_agent import ReActAgent
from droidrun.agent.llm_reasoning import LLMReasoner

async def main():
    # Create an LLM instance
    llm = LLMReasoner(
        llm_provider="gemini",  # "openai", "anthropic", or "gemini"
        model_name="gemini-2.0-flash",
        api_key=os.environ.get("GEMINI_API_KEY"),
        temperature=0.2,
        vision=True  # Enable vision capabilities
    )
    
    # Create and run the agent
    agent = ReActAgent(
        task="Open the Settings app",
        llm=llm,
        vision=True  # Enable screenshot analysis
    )
    
    steps = await agent.run()
    
    # Get token usage statistics
    stats = llm.get_token_usage_stats()
    print(f"Total tokens used: {stats['total_tokens']}")

if __name__ == "__main__":
    asyncio.run(main())

​
Prerequisites

  • Android device connected via USB or ADB over TCP/IP
  • ADB (Android Debug Bridge) installed
  • API key for at least one LLM provider (OpenAI, Anthropic, or Google Gemini)

​
Getting Started

Quickstart

Get up and running with DroidRun in minutes

​
Core Concepts

Agent

Learn about the ReAct agent system

Android Control

Explore Android device interactions