Agent skill

browser-automation

Enterprise-grade browser automation using WebDriver protocol. Use when the user needs to automate web browsers, perform web scraping, test web applications, fill forms, take screenshots, monitor performance, or execute multi-step browser workflows. Supports Chrome, Firefox, and Edge with connection pooling and health management.

Stars 232
Forks 15

Install this agent skill to your Project

npx add-skill https://github.com/aiskillstore/marketplace/tree/main/skills/emillindfors/browser-automation

SKILL.md

Browser Automation Skill

This skill provides guidance for using the rust-browser-mcp server to automate web browsers through the WebDriver protocol. It enables enterprise-grade browser control with performance monitoring, multi-session support, and health management.

Overview

The rust-browser-mcp server provides 45+ MCP tools for browser automation:

Core Automation Tools (25)

  • Navigation: navigate, back, forward, refresh
  • Element Interaction: click, send_keys, hover, find_element, find_elements
  • Information Extraction: get_title, get_text, get_attribute, get_property, get_page_source
  • Advanced: fill_and_submit_form, login_form, scroll_to_element, wait_for_element
  • JavaScript: execute_script
  • Visual: screenshot, resize_window, get_current_url, get_page_load_status

Performance Monitoring Tools (5)

  • get_performance_metrics - Page load times, resource timing, navigation data
  • monitor_memory_usage - Heap monitoring, memory leak detection
  • get_console_logs - Error detection, log filtering
  • run_performance_test - Automated performance analysis
  • monitor_resource_usage - Network, FPS, CPU tracking

Driver Management Tools (7)

  • start_driver, stop_driver, stop_all_drivers
  • list_managed_drivers
  • get_healthy_endpoints, refresh_driver_health
  • force_cleanup_orphaned_processes

Recipe System (4)

  • create_recipe - Create reusable automation workflows
  • execute_recipe - Run a saved recipe
  • list_recipes - List all available recipes
  • delete_recipe - Remove a recipe

Setup Instructions

Prerequisites

Ensure you have at least one WebDriver installed:

  • Chrome: ChromeDriver (must match Chrome version)
  • Firefox: GeckoDriver
  • Edge: MSEdgeDriver

Configuration for Claude Desktop

Add to your claude_desktop_config.json:

json
{
  "mcpServers": {
    "browser": {
      "command": "/path/to/rust-browser-mcp",
      "args": ["--transport", "stdio", "--browser", "chrome"]
    }
  }
}

Environment Variables

Variable Default Description
WEBDRIVER_ENDPOINT auto WebDriver URL or "auto" for auto-discovery
WEBDRIVER_HEADLESS true Run browsers in headless mode
WEBDRIVER_PREFERRED_DRIVER - Preferred browser: chrome, firefox, edge
WEBDRIVER_CONCURRENT_DRIVERS firefox,chrome Browsers to start concurrently
WEBDRIVER_POOL_ENABLED true Enable connection pooling
WEBDRIVER_POOL_MAX_CONNECTIONS 3 Max connections per driver type

Usage Patterns

Basic Navigation

1. Use `navigate` with URL to load a page
2. Use `wait_for_element` to ensure page loads
3. Use `get_title` or `get_text` to verify content

Form Filling

1. Navigate to the form page
2. Use `find_element` with CSS selector to locate fields
3. Use `send_keys` to input values
4. Use `click` on submit button, or use `fill_and_submit_form` for convenience

Web Scraping

1. Navigate to target page
2. Use `find_elements` to get multiple matching elements
3. Use `get_text` or `get_attribute` to extract data
4. Use `execute_script` for complex DOM traversal

Performance Testing

1. Navigate to page under test
2. Use `run_performance_test` for automated analysis
3. Use `get_performance_metrics` for detailed timing data
4. Use `monitor_memory_usage` to detect leaks
5. Use `get_console_logs` to capture errors

Multi-Step Workflows with Recipes

1. Define a recipe with `create_recipe` including steps array
2. Each step specifies: action (tool name), arguments, optional retry logic
3. Execute with `execute_recipe` and parameters
4. Recipes support conditions and browser-specific variants

Session Management

Browser-Specific Sessions

Use session IDs prefixed with browser name for explicit browser control:

  • chrome_session1 - Uses Chrome
  • firefox_work - Uses Firefox
  • edge_testing - Uses Edge

Multi-Session Support

You can run multiple browser sessions concurrently by using different session IDs:

Session: chrome_user1 -> Opens first Chrome tab
Session: chrome_user2 -> Opens second Chrome tab
Session: firefox_admin -> Opens Firefox for different workflow

Best Practices

Error Handling

  1. Always use wait_for_element before interacting with dynamic content
  2. Check get_page_load_status for slow-loading pages
  3. Use get_console_logs to debug JavaScript errors

Performance

  1. Enable connection pooling (default) for better resource usage
  2. Reuse session IDs when possible
  3. Use headless mode for faster execution

Security

  1. Never store credentials in recipes
  2. Use environment variables for sensitive data
  3. Clear sessions after authentication workflows

Troubleshooting

Driver Not Starting

  • Verify WebDriver is installed and in PATH
  • Check browser version matches driver version
  • Use list_managed_drivers to see status

Element Not Found

  • Use browser DevTools to verify selector
  • Wait for page load with wait_for_element
  • Try different selector strategies (CSS, XPath)

Performance Issues

  • Check monitor_memory_usage for leaks
  • Use get_console_logs for JavaScript errors
  • Consider reducing concurrent sessions

Reference Files

See companion files for detailed information:

  • reference/tools.md - Complete tool documentation
  • reference/recipes.md - Recipe system guide
  • examples/ - Example automation scripts

Didn't find tool you were looking for?

Be as detailed as possible for better results