Website Analyzer Agent

An intelligent website analysis tool built with OpenAI Agents SDK and GPT-5 vision that captures website screenshots and provides concise two-sentence summaries of web pages.

Features

• 🖼️ Automated Screenshot Capture: Uses Playwright with headless Chromium
• 🤖 GPT-5 Vision Analysis: Leverages OpenAI's latest vision model for intelligent page analysis
• 🎯 Focused Output: Exactly two sentences summarizing page purpose and key elements
• 🔧 Modular Architecture: Clean separation of tools, agents, and configuration
• 🌐 Cookie Consent Handling: Optional automatic consent button clicking
• ⚙️ Configurable: Customizable viewport sizes, wait conditions, and timeouts

Project Structure

agent_tool/
├── src/
│   ├── agents/
│   │   └── website_analyzer.py    # Main agent implementation
│   ├── tools/
│   │   └── screenshot.py          # Playwright screenshot tool
│   ├── config/
│   │   └── settings.py           # Configuration management
│   └── cli.py                    # CLI utilities
├── tests/
├── main.py                       # Entry point
├── requirements.txt              # Dependencies
├── .env.example                 # Environment template
└── .env                         # Your configuration

Setup

1. Clone and Install Dependencies

cd agent_tool
pip install -r requirements.txt

2. Install Playwright Browsers

playwright install chromium

3. Configure Environment

Copy the example environment file and add your OpenAI API key:

cp .env.example .env

Edit .env and set your OpenAI API key:

OPENAI_API_KEY=your_openai_api_key_here

4. Verify Setup

python main.py --validate-config

Usage

Interactive Mode (Default)

python main.py

Example interaction:

🌐 Website Analyzer Agent
========================================
Powered by OpenAI Agents SDK + GPT-5 Vision

Ready to analyze websites! 🚀
Type 'help' for available commands.

You: Analyze https://stripe.com
🔍 Analyzing...
📝 Stripe is a financial technology platform that enables businesses to accept online payments with a clean, professional design featuring their iconic purple gradient and clear "Start now" call-to-action. The homepage emphasizes trust and simplicity with prominent customer logos, straightforward pricing information, and a streamlined navigation focused on getting started quickly.

Single Query Mode

python main.py --query "Analyze https://example.com"

Available Commands

• help - Show available commands
• config - Display current configuration
• quit/exit - Exit the application

Configuration Options

Edit .env to customize behavior:

# OpenAI Configuration
OPENAI_API_KEY=your_key_here
MODEL_NAME=gpt-5

# Screenshot Settings
DEFAULT_VIEWPORT_WIDTH=1280
DEFAULT_VIEWPORT_HEIGHT=800
DEFAULT_WAIT_UNTIL=networkidle
SCREENSHOT_TIMEOUT=30000

# Agent Settings
MAX_TURNS=3
VERBOSITY=low
REASONING_EFFORT=minimal

Advanced Usage

Custom Screenshot Parameters

The agent automatically handles screenshot capture, but you can provide hints in your queries:

"Analyze https://example.com and wait for the main content to load"
"Take a full page screenshot of https://news-site.com"

Cookie Consent Handling

For sites with cookie banners, you can mention it in your query:

"Analyze https://eu-website.com and handle cookie consent"

API Usage

You can also use the components programmatically:

import asyncio
from src.agents.website_analyzer import WebsiteAnalyzerAgent

async def main():
    agent = WebsiteAnalyzerAgent()
    result = await agent.analyze_website("Analyze https://example.com")
    print(result)

asyncio.run(main())

Development

Running Tests

pytest tests/

Project Structure Explanation

• src/tools/screenshot.py: Playwright-based screenshot capture with async/sync wrappers
• src/agents/website_analyzer.py: OpenAI Agents SDK agent with GPT-5 and custom tools
• src/config/settings.py: Environment-based configuration management
• src/cli.py: Interactive command-line interface utilities
• main.py: Application entry point with argument parsing

Key Design Decisions

1. Modular Architecture: Separate tools, agents, and configuration for maintainability
2. Async/Sync Bridge: Playwright is async, but Agents SDK requires sync functions
3. Temporary Files: Screenshots saved to temp files for GPT-5 vision input
4. Error Handling: Graceful failures with two-sentence error messages
5. Configuration: Environment-based settings for easy deployment

Troubleshooting

Common Issues

1. "Configuration error: OPENAI_API_KEY is required"

   • Set your API key in the .env file

2. "Screenshot capture failed: Navigation timeout"

   • Increase SCREENSHOT_TIMEOUT in .env
   • Site may be blocking automated access

3. "Failed to initialize agent"

   • Verify OpenAI API key has sufficient credits
   • Check if gpt-5 model is available in your region

4. Playwright installation issues

   • Run playwright install chromium
   • On Linux: install additional dependencies with playwright install-deps

Debug Mode

Set environment variable for detailed logging:

export DEBUG=1
python main.py

Contributing

1. Fork the repository
2. Create a feature branch
3. Add tests for new functionality
4. Ensure all tests pass
5. Submit a pull request

License

MIT License - see LICENSE file for details.

Credits

• Built with OpenAI Agents SDK
• Powered by GPT-5
• Screenshot capture via Playwright