Can I use this skill for websites that require authentication?

Yes, Agent Browser includes state save and load commands that allow you to persist login sessions and cookies across different tasks.

Can I run multiple browser sessions at once?

Yes, the skill supports a --session flag which allows you to manage and switch between multiple parallel browser instances.

What is the Agent Browser skill?

Agent Browser is a specialized capability for Claude Code that allows the AI to control a web browser to perform tasks like testing, navigating websites, and extracting data.

Does it support debugging during automation?

Yes, you can use the --headed flag to see the browser window in real-time, or use the console and errors commands to inspect page logs.

How does Agent Browser identify elements on a page?

It uses a snapshot system that generates an accessibility tree and assigns unique references (like @e1 or @e2) to interactive elements, allowing precise interaction without complex CSS selectors.

Agent Browser

Name: Agent Browser
Author: EricOo0

byEricOo0

•

Browser Automation

Automates web browser interactions, testing, and data extraction through a command-line interface for Claude.

About

Agent Browser is a robust automation skill that enables Claude to interact with web applications like a human user. It provides a comprehensive suite of tools for navigating URLs, capturing interactive page snapshots, and performing complex actions such as form filling, clicking, and scrolling using a unique element-referencing system. Designed for high-level tasks, it supports session state management for handling authentication, parallel browsing for multi-site workflows, and various debugging modes including headed browsing and console logging, making it an essential tool for developers building or testing web-based systems.

Key Features

6 GitHub stars
Comprehensive page interactions including clicking, typing, hovering, and scrolling
Interactive snapshot system with unique element references (@refs) for precise control
Support for parallel browser sessions and headless/headed debugging modes
Session state management to save and load cookies/authentication data
High-fidelity screenshot capabilities including full-page and file exports

Use Cases

Automated end-to-end (E2E) testing and web application validation
Automating repetitive web-based tasks like form submissions and account setups
Structured web scraping and data extraction from complex JavaScript-heavy sites

About

Key Features

6 GitHub stars
Comprehensive page interactions including clicking, typing, hovering, and scrolling
Interactive snapshot system with unique element references (@refs) for precise control
Support for parallel browser sessions and headless/headed debugging modes
Session state management to save and load cookies/authentication data
High-fidelity screenshot capabilities including full-page and file exports

Use Cases

Automated end-to-end (E2E) testing and web application validation
Automating repetitive web-based tasks like form submissions and account setups
Structured web scraping and data extraction from complex JavaScript-heavy sites