How does Claude identify elements on a page?

Claude uses the snapshot command to generate an interactive accessibility tree where elements are assigned unique reference tags like @e1 or @e2 for precise targeting.

Can I use this skill for websites requiring login?

Yes, the skill supports saving and loading session states, allowing you to authenticate once and reuse cookies and local storage for future sessions.

Does it support headless and headed modes?

Yes, while it runs headlessly by default, you can use the --headed flag to open a visible browser window for real-time debugging.

Can I take screenshots of the web pages?

Absolutely, the skill includes a screenshot command that can save images to a file or output them directly to stdout for immediate analysis.

Is it possible to run multiple browser sessions at once?

Yes, you can use the --session flag to create and manage multiple independent browser contexts simultaneously for parallel tasks.

Agent Browser Automation

Name: Agent Browser Automation
Author: QuivrHQ

byQuivrHQ

•

Browser Automation

Enables Claude to navigate websites, interact with web elements, and extract data through a controlled browser interface.

The agent-browser skill provides Claude with full web browsing capabilities, allowing it to perform complex tasks such as web application testing, automated form filling, and live data extraction. By utilizing an accessibility tree-based snapshot system with unique reference tags, it ensures precise interaction with DOM elements even on dynamic pages. This skill is essential for developers needing to bridge the gap between their terminal and live web environments, supporting everything from UI verification and end-to-end testing to automated authentication flows and parallel session management.

Key Features

01Session management for parallel browsing and persistent authentication states

02Interactive snapshots with element reference tagging for precise navigation

03Complete DOM interaction suite including click, fill, hover, and scroll

04Semantic locators for finding elements by role, text, or label

055 GitHub stars

06Screenshot capturing with support for full-page and stdout output

Use Cases

01Automated data extraction and web scraping from dynamic JavaScript-heavy sites

02Automating repetitive web-based workflows like form submissions or account setup

03End-to-end testing of web applications directly from the command line

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add quivrhq/247-claude-code-remote agent-browser

For use in Claude.ai and ChatGPT

Download Skill