skills.rest
joneqian
joneqianCOMMUNITY
Explore all skills by this author →
GitHub

agent-browser

Control a headless browser using accessibility-tree snapshots and ref-based element targeting.

1|Updated Mar 12, 2026
Search Tags:#automation#cli#deterministic#refs#headless-browser#agent-browser#accessibility-tree
One-click install
npx skills add https://github.com/joneqian/EvoClaw --skill agent-browser-joneqian
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/joneqian/EvoClaw/tree/main/packages/core/src/skill/bundled/agent-browser-clawdbot
Command: npx skills add https://github.com/joneqian/EvoClaw --skill agent-browser-joneqian

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Headless browser automation optimized for AI agents, offering deterministic element targeting via ref-based selectors and accessibility-tree snapshots to simplify complex web interactions.

Core Features & Use Cases

  • Deterministic element selection using refs for reliable automation on dynamic SPAs.
  • Session isolation and powerful navigation, scripting, and state persistence.
  • JSON-friendly snapshot and interaction outputs to feed AI agents.

Quick Start

Open a URL, snapshot the page with interactive elements, then drive actions using refs and export JSON.

Frequently Asked Questions about agent-browser

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

FAQPage Schema
How do I automate deterministic web interactions on dynamic SPAs using AI agents?▼

Deterministic web interactions on dynamic SPAs are automated by controlling a headless browser with accessibility-tree snapshots and ref-based element targeting. This ensures precise, reliable execution for multi-step AI-assisted workflows.

What is ref-based element targeting in headless browser automation?▼

Ref-based element targeting in headless browser automation uses stable references derived from accessibility-tree snapshots to interact with elements. This approach maintains consistent targeting on complex single-page applications where DOM structures change dynamically.

How do I get JSON-friendly outputs from a headless browser for AI processing?▼

You get JSON-friendly outputs from a headless browser by taking accessibility-tree snapshots of interactive elements and exporting the interaction results. These structured outputs feed directly into AI agents for state evaluation and action planning.

Does headless browser automation support session isolation for parallel AI workflows?▼

Headless browser automation supports session isolation to manage parallel AI workflows securely. This feature separates browser contexts, preventing state interference and ensuring precise execution across independent multi-step tasks.

What's the best way to drive actions on a web page using accessibility-tree snapshots?▼

The best way to drive actions on a web page using accessibility-tree snapshots is through a CLI tool workflow. You open a URL, snapshot interactive elements, target them via refs, execute actions, and export the results as JSON.

Related Skills

Explore More Software Engineering Skills→
ronakpjainronakpjain
10

playwright-firefox

Automate Firefox browser sessions for live, headless, and multi-profile testing.

Community
Intermediate
pekralpekral

autonomous-loops

Orchestrate autonomous AI agent loops with quality gates and stop conditions.

Community
Advanced
liuaibin001liuaibin001

automating-browsers

Automate browser interactions for AI agents using the agent-browser CLI.

Community
Advanced
libilibi
27

ko-browser

Automate browser navigation, form submission, and screenshot capture via a Go-based CLI.

Community
Advanced