gstack

Run a persistent Playwright headless browser daemon for Claude Code workflows.

Updated Mar 19, 2026
One-click install
npx skills add https://github.com/him55710-sudo/uni-foli --skill gstack-him55710-sudo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gstack
Source: https://github.com/him55710-sudo/uni-foli/tree/main/.agents/skills/gstack
Command: npx skills add https://github.com/him55710-sudo/uni-foli --skill gstack-him55710-sudo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

gstack provides a fast, persistent browser daemon for Claude Code skills, enabling reliable QA testing, design reviews, and bug hunting inside AI workflows without rebuilding the browser state each time.

Core Features & Use Cases

  • Persistent headless browser powered by Playwright for deterministic interactions across skills like /qa, /design-review, and /browse.
  • Ref-based accessibility snapshot system with stable @e and @c references for reliable element interaction.
  • State persistence across commands (cookies, tabs, localStorage where supported) and idle shutdown to conserve resources.
  • Secure, token-authenticated control over the browser with safe defaults and automatic crash/recovery handling.

Quick Start

Install gstack, run the setup, and start a /browse session to begin QA dogfooding in a local browser.

Frequently Asked Questions about gstack

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

FAQPage Schema
How do I run headless browser QA testing inside Claude Code workflows?

You can run QA testing by installing the daemon and starting a /browse session, which provides a persistent headless browser powered by Playwright for deterministic interactions and state persistence across commands.

What is a persistent headless browser daemon and when do I need one for AI sessions?

A persistent headless browser daemon maintains browser state across multiple AI commands, needed when performing continuous QA testing, design reviews, or bug hunting inside AI workflows without losing session state or cookies.

Can I use Playwright for deterministic browser interactions in AI automation?

Yes, Playwright powers the daemon and uses a ref-based accessibility snapshot system with stable references, ensuring deterministic element interactions that remain reliable across automated testing and dogfooding sessions.

Does the headless browser daemon persist cookies and localStorage across commands?

Yes, the headless browser daemon persists state including cookies, open tabs, and localStorage where supported, allowing sequential QA commands to maintain context without restarting or rebuilding the browser environment.

How do I handle browser crashes during automated QA testing workflows?

The daemon includes automatic crash and recovery handling, ensuring browser failures during QA testing workflows are managed securely with safe defaults and idempotent operations to maintain testing continuity.

What are the limitations of using a headless browser for design reviews in AI sessions?

Headless browsers lack visual rendering, so design reviews in AI sessions rely on accessibility snapshots and ref-based element references rather than visual inspection, requiring token-authenticated control for secure operations.