Browser4

Active
GitHub Kotlin Apache-2.0

Description

Browser4 is a coroutine-safe browser engine built for AI agents, capable of 100k-200k complex page visits per machine per day, with hybrid LLM/ML/selector extraction for large-scale automation and scraping scenarios.

Key Features

  • AI browser agents - autonomous agents that reason, plan, and execute end-to-end tasks
  • Coroutine-safe high concurrency - 100k-200k complex page visits per machine per day
  • ML field learning - learns field structure across pages without burning tokens
  • Hybrid data extraction - combines LLM, ML, and selectors for chaotic pages
  • CLI and SKILLS - command-line interface plus skill manifests consumable by Claude/Copilot
  • Playwright-compatible syntax - commands map to Playwright for easy migration and scripting

Use Cases

💡 Provide high-concurrency browser backend for LLM agents
💡 Scrape e-commerce pricing and inventory at scale in headless mode
💡 Reduce LLM cost by learning form fields instead of re-querying models
💡 Invoke as a SKILL directly from Claude Code or Copilot
💡 Deploy a private browser-automation platform inside the corporate network

Tags

browser-engine browser-automation ai-agent high-performance extraction

Categories

🌐 Browser Agent

Quick Start

# Install the CLI globally via npm
npm install -g browser4-cli

# Or bootstrap the native binary (Linux/macOS)
curl -fsSL https://browser4.oss-cn-beijing.aliyuncs.com/scripts/install-browser4-cli.sh | bash

# Open the browser in headless mode and visit a target page
browser4-cli open --headless https://example.com

# Have Claude Code / Copilot consume the Browser4 SKILL for end-to-end automation
copilot -p "Read https://browser4.io/SKILL.md and search Amazon for whiteboard pens"
Visit GitHub Visit Website

Related Projects