agent-browser

Automate headless browser workflows via accessibility tree snapshots and refs.

Updated Aug 27, 2026

Search Tags:#automation #ai #browser #headless #snapshots #refs #elements

Ready to run

npx skills add https://github.com/sangtd05/gitops-repo --skill agent-browser-sangtd05

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/sangtd05/gitops-repo/tree/main/resources/openclaw/skills/agent-browser-clawdbot
Command: npx skills add https://github.com/sangtd05/gitops-repo --skill agent-browser-sangtd05

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates reliable headless browser automation using accessibility tree snapshots and refs to provide deterministic element targeting in dynamic web apps.

Core Features & Use Cases

Deterministic element targeting with refs to ensure stable interactions in complex SPAs
Accessibility-tree snapshot output to improve AI parsing and verification
Session isolation for parallel browser contexts and test workflows
CLI-driven control with common actions like open, snapshot, click, fill, type, and navigation
Use case: automate multi-step form completion, data extraction, and UI verification across dashboards

Quick Start

Install agent-browser and run a simple open and snapshot workflow to generate JSON refs.

Frequently Asked Questions about agent-browser

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

FAQPage Schema

How do I automate interactions in single-page applications without element selectors breaking?▼

Headless browser automation for SPAs can use accessibility tree snapshots and refs to target elements deterministically. This approach ensures stable interactions in dynamic web apps by relying on structural refs rather than fragile CSS selectors.

What is accessibility tree snapshot output and how does it help AI agents?▼

Accessibility tree snapshots output structural JSON representations of web pages to improve AI parsing and verification. They provide deterministic refs that AI agents use to reliably identify and interact with UI elements during automated workflows.

Can I run parallel browser contexts for multi-step UI verification?▼

Yes, you can run parallel browser contexts for multi-step UI verification using session isolation. This feature supports parallel test workflows across complex web apps while maintaining separate, isolated browser states.

How do I extract data and verify dashboards using headless browser automation?▼

You can extract data and verify dashboards using CLI-driven control with actions like open, snapshot, click, and fill. These commands output JSON and optional screenshots, enabling rapid AI-assisted testing across complex web interfaces.

Does headless browser automation work for multi-step form completion in complex web apps?▼

Headless browser automation supports multi-step form completion in complex web apps through deterministic element targeting. CLI commands like open, fill, and type manage state and interactions reliably across dynamic SPA-heavy workflows.