Can I use this for authenticated sessions?

Yes, the skill supports saving and loading session states, including cookies and localStorage, into JSON files which can be encrypted to maintain persistent logins across different sessions.

Does it support mobile browser testing?

Absolutely. It includes dedicated platform flags for iOS, allowing the agent to control the iOS Simulator and Mobile Safari for mobile-specific interactions and gestures.

What is 'Vision Mode' in this skill?

Vision Mode uses the --annotate flag to generate screenshots where interactive elements are overlaid with numbered labels, helping the AI model reason about visual layouts and icon-only buttons.

Is chaining commands supported for efficiency?

Yes, multiple browser commands can be chained using '&&' in a single shell invocation. Because the browser runs via a background daemon, the state is preserved between these calls.

How does the skill handle dynamic content?

It uses a snapshot-and-interact pattern where Claude takes a DOM snapshot to generate stable element IDs (@refs), performs an action, and then re-snapshots to capture any dynamic updates or navigation changes.

Agent Browser Automation

Name: Agent Browser Automation
Author: Jamie-BitFlight

byJamie-BitFlight

•

浏览器自动化

Automates web browser interactions, data extraction, and testing for AI agents through a structured CLI-driven workflow.

This skill empowers Claude to interact with the web directly through a powerful CLI-based browser automation suite. It allows agents to navigate complex websites, fill forms, handle authentication with state persistence, and capture screenshots including annotated vision-mode views for spatial reasoning. By providing a structured 'snapshot-and-interact' pattern with stable element references, it enables reliable web scraping, automated end-to-end testing, and multi-step browser workflows across both desktop and iOS mobile environments. It is particularly useful for tasks requiring high precision in dynamic, JavaScript-heavy web applications.

主要功能

01Annotated visual screenshots that map interactive elements to AI-readable labels

02Session management with encrypted state persistence for persistent logins

0328 GitHub stars

04Interactive DOM snapshots with stable element references (@e1, @e2) for precise targeting

05Cross-platform support including iOS Simulator and Mobile Safari automation

06Advanced DOM and visual regression diffing to verify page changes after actions

使用场景

01Automating repetitive browser workflows like form submissions, account setups, or CMS management

02Automated end-to-end testing and visual regression checks for web applications

03Complex data extraction and web scraping from dynamic, authenticated sites

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jamie-bitflight/claude_skills agent-browser

For use in Claude.ai and ChatGPT

Download Skill