browser-agent

Name: browser-agent
Availability: InStock
Author: tom-abp

Automate browser UI verification using accessibility tree snapshots via agent-browser CLI.

Updated Sep 30, 2025

Search Tags:#automation #accessibility #qa #ui-testing #browser

Ready to run

npx skills add https://github.com/tom-abp/.claude --skill browser-agent-tom-abp

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: browser-agent
Source: https://github.com/tom-abp/.claude/tree/main/skills/browser-agent
Command: npx skills add https://github.com/tom-abp/.claude --skill browser-agent-tom-abp

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automate browser-based UI verification and feature exploration, reducing manual testing time and accelerating feedback loops.

Core Features & Use Cases

CLI-based browser automation using agent-browser (Vercel Labs). Uses accessibility tree snapshots — no screenshots, no vision model, token-efficient.
Facilitates end-to-end UI testing, accessibility checks, and quick feature discovery across running web apps.
Supports structured handoffs with a goal, url, context, and optional acceptance criteria to guide automated exploration.

Quick Start

Install agent-browser globally, run agent-browser install, and load the browser-agent skill reference to begin automated UI testing.

Frequently Asked Questions about browser-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I automate browser UI testing for acceptance criteria?▼

Automate browser UI testing using accessibility tree snapshots instead of screenshots or vision models. This approach provides token-efficient, deterministic automation for verifying UI journeys without relying on visual processing.

Do I need agent-browser installed to run browser automation tasks?▼

Yes, agent-browser CLI is required. Install it globally, run agent-browser install, and load the browser-agent skill reference to start automating browser-based UI verification and feature exploration.

How does accessibility tree-based UI testing work without screenshots?▼

Accessibility tree-based UI testing uses structured snapshots of the web page's accessibility tree instead of screenshots. This avoids vision models, ensuring deterministic, token-efficient automation for verifying UI journeys.

Can I verify feature exploration on running web apps with CLI browser automation?▼

Yes, CLI browser automation supports feature exploration and end-to-end UI testing on running web apps. You provide a structured handoff including a goal, URL, context, and acceptance criteria to guide automated verification.

What context do I need to provide for automated end-to-end UI journeys?▼

You need a structured handoff payload containing the testing goal, target URL, application context, and optional acceptance criteria. This payload guides the agent-browser CLI through deterministic, snapshot-first interactions.

Does snapshot-first browser testing support accessibility checks?▼

Yes, snapshot-first browser testing uses accessibility tree snapshots which inherently support accessibility checks. By interacting with the accessibility tree, it facilitates both end-to-end UI testing and accessibility verification across running web applications.