browser-use Skill for Claude Code

A Claude Code skill that wraps the official browser-use library, enabling AI-powered browser automation through two modes:

1. Direct Mode - Claude directly controls browser via Actor API (no external LLM API key required!)
2. Subagent Mode - Delegate complex tasks to autonomous Claude Code subagents

Why This Skill?

The official browser-use library is designed for standalone Python scripts with LLM integration. This skill adapts it for Claude Code by:

• Server Mode: Maintains browser session across multiple tool calls
• Direct Control: Claude can control browser step-by-step using Vision
• No API Key Required: Direct Mode works without OpenAI/Gemini API keys
• Session Persistence: Browser stays open until explicitly closed

Features

• Navigate to any URL
• Click elements by index or coordinates
• Type text into forms
• Take screenshots (Vision-compatible)
• Keyboard input and shortcuts
• Scroll and mouse actions
• Tab management
• JavaScript evaluation

Installation

1. Install dependencies

pip install browser-use
playwright install chromium

2. Copy skill to Claude Code

# Clone this repo
git clone https://github.com/tau-breath/browser-use-skill.git

# Copy to Claude Code skills directory
cp -r browser-use-skill ~/.claude/skills/browser-use

3. Verify installation

cd ~/.claude/skills/browser-use
python server.py start &
sleep 2
python server.py status
python server.py stop

Quick Start

Direct Mode (No API Key Required!)

cd ~/.claude/skills/browser-use

# Start server (keep running in background)
python server.py start &
sleep 2

# Navigate
python server.py call '{"tool": "navigate", "args": {"url": "https://google.com"}}'

# Get page state with screenshot
python server.py call '{"tool": "get_state", "args": {"include_screenshot": true}}'

# Type into search box (index from get_state)
python server.py call '{"tool": "type", "args": {"index": 0, "text": "Claude AI"}}'

# Press Enter
python server.py call '{"tool": "press_key", "args": {"key": "Enter"}}'

# Screenshot results
python server.py call '{"tool": "screenshot", "args": {"path": "results.png"}}'

# Stop server when done
python server.py stop

Server Commands

Command	Description
python server.py start &	Start server in background
python server.py stop	Stop server
python server.py status	Check server status
python server.py call '{...}'	Call a tool

Available Tools

Page Tools

• navigate - Go to URL
• go_back / go_forward - Navigation history
• reload - Refresh page
• get_state - Get elements + optional screenshot
• screenshot - Save screenshot to file
• evaluate - Run JavaScript
• press_key - Keyboard input

Element Tools

• find_elements - Find by CSS selector
• click - Click element by index
• type - Type text into element
• hover - Mouse hover
• check - Toggle checkbox
• select_option - Select dropdown option
• drag_to - Drag and drop

Mouse Tools

• mouse_click - Click at coordinates
• mouse_move - Move mouse
• mouse_drag - Drag from A to B
• scroll - Scroll page

Tab Management

• list_tabs - List open tabs
• switch_tab - Switch to tab
• close_tab - Close tab
• close - Close browser

Example: Search Naver (Korean Portal)

# Start server
cd ~/.claude/skills/browser-use && python server.py start &
sleep 2

# Navigate to Naver search
python server.py call '{"tool": "navigate", "args": {"url": "https://search.naver.com/search.naver?query=Claude+AI"}}'

# Get screenshot
python server.py call '{"tool": "get_state", "args": {"include_screenshot": true}}'
# Returns: {"url": "...", "elements": [...], "screenshot_path": "/path/to/screenshot.png"}

# Read screenshot with Vision to analyze results

Troubleshooting

Server not running

python server.py start &
sleep 2

Element click not working

1. Use get_state to refresh element cache
2. Try press_key("Tab") then press_key("Enter")
3. Use mouse_click(x, y) with coordinates from screenshot

Session lost

Always use server.py commands. Direct Python calls don't maintain session.

How It Works

Without Server Mode:
  Call 1: navigate -> opens browser -> closes on exit (state lost!)
  Call 2: click -> ERROR: no browser!

With Server Mode:
  Start server: browser session created
  Call 1: navigate -> works
  Call 2: click -> works (same session!)
  Call N: ... -> works
  Stop server: browser closes

Requirements

• Python 3.8+
• browser-use (pip install browser-use)
• Playwright + Chromium (playwright install chromium)
• (Optional) LLM API key for run_agent tool

License

MIT

Credits

• browser-use - The official browser automation library

---

DPA - Decentralized Protection Alliance

Freedom without surveillance, protection for everyone.

dpa.network | contact@dpa.network | Matrix | Telegram