Is this approach more maintainable than XPath?

Yes, accessibility tree references are generally more stable and meaningful than CSS selectors or XPaths, which often break when the underlying HTML structure changes slightly.

Why use the accessibility tree instead of screenshots?

Screenshots are prone to failure with UI changes, different screen sizes, or themes, while the accessibility tree provides stable semantic references that remain consistent.

Does this skill work with interactive command-line tools?

Yes, it includes guidance for using tmux to manage, send commands to, and capture output from interactive CLI sessions effectively.

How does the skill locate specific buttons or inputs?

It uses the find command to search the page structure for specific descriptions like 'Submit button' or 'Email input field' rather than using fragile coordinates.

Chrome Browser Automation

Name: Chrome Browser Automation
Author: igrybkov

byigrybkov

0•

Browser Automation

Enables reliable Chrome browser interaction by leveraging accessibility tree references instead of visual screenshots.

This skill empowers Claude to navigate and interact with the Google Chrome browser using a semantic, tree-based approach. By utilizing the accessibility tree via the read_page and find commands, it identifies elements by their functional roles and descriptions rather than fragile coordinates or visual screenshots. This method ensures high reliability across different screen sizes, themes, and minor UI updates, making it an essential tool for robust web-based tasks, automated browser control, and managing interactive command-line interfaces through tmux integration.

Key Features

010 GitHub stars

02Semantic element interaction using accessibility tree references

03Real-time page state verification for accurate task execution

04Integrated tmux support for managing interactive CLI sessions

05Advanced element discovery through descriptive text matching

06Reliable page navigation without dependency on visual screenshots

Use Cases

01Automating complex multi-step web workflows and form submissions

02Extracting data from dynamic web applications with stable semantic references

03Performing cross-platform web testing where visual layouts may vary

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add igrybkov/dotfiles claude-for-chrome

For use in Claude.ai and ChatGPT

Key Features

010 GitHub stars

02Semantic element interaction using accessibility tree references

03Real-time page state verification for accurate task execution

04Integrated tmux support for managing interactive CLI sessions

05Advanced element discovery through descriptive text matching

06Reliable page navigation without dependency on visual screenshots

Use Cases

01Automating complex multi-step web workflows and form submissions

02Extracting data from dynamic web applications with stable semantic references

03Performing cross-platform web testing where visual layouts may vary

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add igrybkov/dotfiles claude-for-chrome

For use in Claude.ai and ChatGPT