Web Browser AI-Control API Server

A hosted API server for AI agents to control web browsers in the cloud using Browserbase, Stagehand, and Playwright.

Overview

This server provides a REST API for AI agents to:

• Create and manage browser sessions in the cloud
• Control browsers using natural language commands
• Extract structured data from web pages
• Monitor browser activity and errors
• Take screenshots and inspect DOM state
• Execute custom Playwright automation scripts

Built on top of:

• Browserbase for cloud browser infrastructure
• Stagehand for AI-powered browser control
• Playwright for low-level browser automation
• Express.js for the API server

API Reference

For complete API documentation, visit wbac-api-docs.netlify.app

Session Management

• POST /api/sessions/create-session - Create new browser session
• POST /api/sessions/stop-session/:sessionId - Stop session
• GET /api/sessions/running-sessions - List active sessions
• GET /api/sessions/session/:sessionId - Get session information
• GET /api/sessions/debug/:sessionId - Get session debug URLs

Browser Control

• POST /api/browser/navigate/:sessionId - Navigate to URL
• POST /api/browser/act/:sessionId - Perform action via natural language
• POST /api/browser/extract/:sessionId - Extract structured data
• POST /api/browser/observe/:sessionId - Get possible actions

Monitoring

• GET /api/browser/console-logs/:sessionId - Get console logs
• GET /api/browser/network-logs/:sessionId - Get network logs
• GET /api/browser/dom-state/:sessionId - Get DOM state
• POST /api/browser/screenshot/:sessionId - Take screenshot
• POST /api/browser/clear-logs/:sessionId - Clear logs

Key Features

Browser Session Management

• Create new browser sessions
• Resume existing sessions
• List running sessions
• Stop/cleanup sessions

AI-Powered Browser Control

• Natural language actions via act()
• Structured data extraction via extract()
• Page observation via observe()
• Vision-based interaction support

Monitoring & Debugging

• Console log monitoring
• Network request/response logging
• Error tracking
• Screenshot capture
• DOM state inspection

Getting Started

Prerequisites

• Node.js 16+
• Browserbase account and credentials
• OpenAI or Anthropic API key for AI features

Installation

git clone https://github.com/kortix-ai/wbac
cd wbac
npm i

Configuration

Create a .env file with:

BROWSERBASE_API_KEY=your_api_key
BROWSERBASE_PROJECT_ID=your_project_id
OPENAI_API_KEY=your_openai_key 
ANTHROPIC_API_KEY=your_anthropic_key

Running the Server

npm start

UI Interface

A Streamlit-based UI is included for testing and debugging:

pip install streamlit
streamlit run streamlit_ui.py

Use Cases

• AI agents that need web browsing capabilities
• Automated web testing with AI assistance
• Web scraping with natural language commands
• Browser automation monitoring and debugging

Contributing

Contributions welcome! Please read our contributing guidelines and submit pull requests.

License

MIT License - see LICENSE file for details

Acknowledgements

• Browserbase for cloud browser infrastructure
• Stagehand for AI browser control capabilities
• Playwright for powerful browser automation