browse

Controls a persistent headless Chromium browser for QA testing and site verification.

Updated Apr 5, 2026
One-click install
npx skills add https://github.com/VertaKhan/cs_gstack --skill browse-vertakhan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: browse
Source: https://github.com/VertaKhan/cs_gstack/tree/main/.claude/skills/gstack/browse
Command: npx skills add https://github.com/VertaKhan/cs_gstack --skill browse-vertakhan

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires playwright, playwright-core, diff, bun, and includes scripts (resource) components.

What problem does it solve? Manually verifying that a web feature works, a deployment renders correctly, or a user flow completes requires opening a browser, clicking through pages, and capturing evidence. This Skill automates that loop with a persistent headless Chromium instance, letting an AI agent navigate URLs, interact with elements, assert page state, and produce screenshots as bug evidence at roughly 100ms per command. ## Core Features & Use Cases - Page Verification: Navigate any URL and check text content, console errors, failed network requests, and element visibility in one pass. - User Flow Testing: Fill forms, click buttons, upload files, handle dialogs, and diff accessibility snapshots before and after actions to confirm exactly what changed. - Visual Evidence: Capture plain or annotated screenshots with labeled element refs, plus responsive screenshots at mobile, tablet, and desktop viewports for bug reports. - User Handoff: When blocked by CAPTCHAs or multi-factor auth, hand off to a visible Chrome window preserving all cookies and session state, then resume control. - Use Case: After deploying a login flow, run goto, fill credentials, click submit, then snapshot -D to diff the result and assert the dashboard element is visible. ## Quick Start Ask the agent to open your staging site in the browser, click through the signup flow, and take an annotated screenshot of the result.

Frequently Asked Questions about browse

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

FAQPage Schema
How do I test a website with a headless browser from the command line?▼

Use the browse CLI to navigate with goto, inspect content with text and console, and interact via click and fill commands. The persistent server keeps cookies and tabs between calls, so multi-step flows run at about 100ms per command after startup.

How to verify a user flow works after a deployment?▼

Navigate to the page, run snapshot -i to list interactive elements, fill and click through the flow using @e refs, then run snapshot -D to diff what changed. Finish with an is visible assertion on the expected success element.

Can a headless browser handle CAPTCHAs or multi-factor login?▼

Headless automation cannot solve CAPTCHAs or MFA. The handoff command opens a visible Chrome window at the current page preserving all cookies and state, the user completes the challenge, and resume returns control to the agent.

Does the browse tool work on Windows?▼

Yes, but Bun cannot launch Playwright Chromium on Windows, so the server falls back to a Node.js bundle with Bun API polyfills. The build-node-server.sh script generates server-node.mjs for this compatibility path.

Why do element refs stop working after navigation?▼

Refs like @e3 point to live element locators that become stale when the page navigates. Run snapshot again after any goto or page change to get fresh refs before interacting.