AI Browser Agent

An autonomous AI-powered browser that can take control of your browser, perform tasks, and remember what it does.

Features

• Browser Automation: Control browser actions (navigation, clicking, filling forms, etc.)
• Memory System: Persist and retrieve information about previous interactions
• Multi-AI Support: Works with any AI provider (OpenAI, Anthropic, or custom)
• Task Execution: Autonomous task execution based on natural language instructions
• Action History: Complete audit trail of all browser interactions

Architecture

src/
  ├── browser/       # Browser control & automation
  ├── memory/        # Memory & context management
  ├── agents/        # AI agent interfaces & implementations
  └── core.py        # Main orchestrator
config/
  └── config.yaml    # Configuration file
tests/              # Test files

Installation

# Install dependencies
pip install -r requirements.txt

# Install Playwright browsers
python -m playwright install

Quick Start

from src.core import AIBrowser

# Initialize browser
browser = AIBrowser(ai_provider="openai")

# Execute a task
result = await browser.execute_task("Find the price of Python books on Amazon")

Configuration

Edit config/config.yaml to configure:

• AI Provider (OpenAI, Anthropic, etc.)
• Memory settings
• Browser headless mode
• Task execution parameters

Supported AI Providers

• OpenAI (GPT-4, GPT-3.5-turbo)
• Anthropic (Claude)
• Custom providers (implement the BaseAIAgent interface)