browser-ui-review

Inspect and interact with live web pages in a real browser to verify UI behavior and capture screenshot evidence.

5|1|Updated May 5, 2026
One-click install
npx skills add https://github.com/cbusillo/codex-skills --skill browser-ui-review-cbusillo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: browser-ui-review
Source: https://github.com/cbusillo/codex-skills/tree/main/skills/browser-ui-review
Command: npx skills add https://github.com/cbusillo/codex-skills --skill browser-ui-review-cbusillo

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Verifying web UI behavior from source code or static HTML alone leads to missed defects. This Skill drives a real browser session to open pages, click, type, scroll, and inspect visible state, so UI claims are validated against what users actually see. ## Core Features & Use Cases - Live Browser Interaction: Open pages, click, fill forms, scroll, and inspect visible state using the host's browser capability or an installed ui-browser CLI. - Structured UI QA Inventory: Build a checklist of claims, controls, states, and viewports before signoff, covering both functional and visual checks. - Design Validation: Compare the live result against accepted design direction from GitHub issues or design notes, classifying differences as tradeoffs, gaps, or quality issues. - Use Case: After implementing a settings page, use this Skill to open it in a browser, submit the form, verify the success and error states at desktop and mobile widths, and capture screenshots as PR evidence. ## Quick Start Open the local dev server page in a browser, click through the main user flow, verify the visible states, and save screenshots to scratch/ui-checks/.

Frequently Asked Questions about browser-ui-review

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

FAQPage Schema
How do I verify a web UI in a real browser instead of reading the code?

Open the page with the host's browser capability or the ui-browser CLI, take a snapshot to ground selectors in live state, then click, fill, and scroll like a user. Re-snapshot after each navigation or major UI change and capture screenshots as evidence.

How do I test a web page with the ui-browser CLI?

Start a named session with ui-browser open, wait for readiness with wait-for, then use snapshot, click, fill, and scroll commands. Keep the same --session flag across commands and close only the session you created when finished.

Can I attach to an existing Chrome or Edge tab for UI testing?

Yes, if the host exposes a computer-use API or browser extension that can attach to that tab. A new ui-browser session does not share the existing tab's profile, cookies, or authentication state, so report the limitation if no attach capability exists.

Why do my browser selectors become stale during UI automation?

Selectors go stale when navigation, modals, or form submissions change the DOM. Recover by taking a fresh snapshot and choosing the next selector from the new visible state instead of forcing the old reference through JavaScript eval.

When should I avoid screenshot-only UI verification?

A plausible screenshot does not prove controls work, and a passing click does not prove the layout is acceptable. Cover both functional interaction and visual inspection, using screenshots as supporting evidence rather than the whole workflow.