browser

djgrant's avatarfrom djgrant

When testing or verifying web UI use this skill to start the browser server, take screenshots to observe state, interact with elements, and verify visually.

⭐2stars🔀0forks📁View on GitHub🕐Updated Jan 8, 2026⬇️2downloads

When & Why to Use This Skill

This Claude skill enables automated web UI testing and browser interaction, allowing agents to programmatically navigate websites, capture screenshots for visual verification, and interact with DOM elements. It is an essential tool for ensuring web interface integrity and automating repetitive browser-based tasks.

Use Cases

  • Automated UI Testing: Systematically verify web components and layouts by capturing screenshots and comparing them against expected states.
  • End-to-End Functional Validation: Simulate complex user journeys by programmatically clicking elements and typing text to ensure workflows function correctly.
  • Visual Regression Monitoring: Use visual snapshots to detect unintended UI changes or rendering issues during the development lifecycle.
  • Dynamic Web Interaction: Navigate through multi-page web applications to extract data or perform actions within state-dependent environments.
namebrowser
descriptionWhen testing web UI, load this skill to take screenshots and interact with elements.

::workflow::

DO

  1. open to start session
  2. snapshot to see current state and get @refs
  3. interact with elements (click, type, etc.) using @refs or selectors
  4. screenshot to verify result END

Commands

agent-browser <command>

Common commands: open <url>, snapshot, click <selector|@ref>, type <selector|@ref> <text>, screenshot. Use agent-browser --help for full command list.