browser

Automate web browsing and UI interactions using accessibility-tree snapshots with stable element refs.

Updated Apr 12, 2026
One-click install
npx skills add https://github.com/softmg/product-tracker --skill browser-softmg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: browser
Source: https://github.com/softmg/product-tracker/tree/main/.claude/skills/browser
Command: npx skills add https://github.com/softmg/product-tracker --skill browser-softmg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Web pages contain large, noisy DOMs that make automated scraping and interaction brittle and expensive for AI agents. This Skill reduces context size by using AI-optimized accessibility-tree snapshots and stable element references so agents can navigate, fill forms, click controls, and extract data reliably.

Core Features & Use Cases

  • AI-optimized snapshots with element refs: produce compact accessibility trees that replace full DOM dumps with stable refs (e.g., @e1) for deterministic actions.
  • Full browser control: navigation (open, back, forward, reload, close), interaction (click, fill, type, press, hover, select, check), and information extraction (get text, html, value, attrs, title, url).
  • Session and workflow management: isolated sessions for parallel tasks, session state save/load, multi-session swarm workflows for coordinated scraping or testing.
  • Integration and automation: screenshot support, snapshot flags for interactive-only or compact output, MCP tool namespacing (browser/*), memory storage and hooks for pre/post task orchestration.
  • Use cases: automated login and form submission, product data extraction from listings, end-to-end UI smoke tests, and authenticated multi-session scraping.

Quick Start

Open a URL, take an interactive snapshot, and extract product names and prices using the element refs from the snapshot.

Frequently Asked Questions about browser

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

FAQPage Schema
How do I automate web scraping with AI agents without hitting large DOM context limits?

You can automate web scraping using AI-optimized accessibility-tree snapshots with stable element refs. This replaces full DOM dumps with compact outputs, reducing context size and making automated scraping reliable for AI agents.

What's the best way to extract product data from web listings using AI browser automation?

The best way to extract product data is taking an interactive snapshot of the page and using the generated stable element references to click controls and get text attributes. This AI browser automation method ensures deterministic data extraction.

Can I run multiple browser sessions in parallel for authenticated scraping tasks?

Yes, you can run isolated browser sessions for parallel tasks. The skill supports session state save/load and multi-session swarm workflows, enabling coordinated authenticated scraping across multiple isolated sessions.

How do accessibility-tree snapshots improve automated UI testing compared to full DOM dumps?

Accessibility-tree snapshots improve automated UI testing by providing compact, interactive-only outputs with deterministic element refs instead of noisy full DOM dumps. This reduces context size and prevents brittle, expensive UI test execution.

Does claude-flow MCP integration support browser navigation and form filling automation?

Yes, claude-flow MCP integration supports browser automation including navigation, form filling, and clicks. It uses MCP tool namespacing with memory storage and hooks for pre/post task orchestration during web interactions.

What browser interactions are supported for automated login and form submission?

Supported browser interactions for automated login include click, fill, type, press, hover, select, and check. These controls work with stable element refs from snapshots to complete form submission tasks reliably.