ux-review

Walks a live UI as a first-time user and reports usability friction with severity ratings.

Updated May 12, 2026
One-click install
npx skills add https://github.com/fryanpan/ai-team-lead --skill ux-review-fryanpan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ux-review
Source: https://github.com/fryanpan/ai-team-lead/tree/main/plugin/team-lead-fleet/skills/ux-review
Command: npx skills add https://github.com/fryanpan/ai-team-lead --skill ux-review-fryanpan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? UI features often ship after only being reviewed in code, so real users hit confusion, dead-ends, and wrong-target clicks in production and send the work back. This Skill makes the agent actually use the page in a browser before it is called done, catching friction before the user does. ## Core Features & Use Cases - Cold walkthrough of user goals: Lands on the page fresh, attempts each user goal, tries to break the flow, and tests at mobile width (~375px). - Heuristic evaluation: Scores the page against Nielsen's 10 heuristics, Fitts's Law target sizing, visual hierarchy, and goal completion with Pass / Issue / Critical ratings. - Severity-ranked report: Produces a markdown report grouping findings into Critical (block ship), Issues (fix before PR), and Polish (post-ship), with screenshots at key states. - Use Case: After implementing a new checkout flow, run this Skill against the local dev server before opening the PR to catch ambiguous labels, missing feedback, and broken mobile layouts. ## Quick Start Run a UX review of the new settings page at http://localhost:3000/settings, walking the goal of updating a notification preference, and report any critical issues before I open the PR.

Frequently Asked Questions about ux-review

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

FAQPage Schema
How do I review a UI feature for usability before shipping?▼

Run the UI locally or use a deployed URL, then walk each user goal in a real browser as a first-time user. Evaluate the page against Nielsen's heuristics, Fitts's Law target sizes, and goal completion, then report findings grouped by severity: Critical, Issues, and Polish.

What tools are needed to run a browser-based UX review?▼

This workflow requires the claude-in-chrome MCP tools connected to a running Chrome instance, plus a live dev server or deployed URL. If the browser connection is unavailable, the review must stop rather than falling back to reading the code.

Can I review usability from the code instead of running the UI?▼

No. Reviewing from code defeats the purpose because the goal is to see what a real user sees. If the UI cannot be run in a browser, the correct action is to stop and say so rather than grade the implementation statically.

Should the agent that built the feature do the UX review?▼

No. Familiarity hides friction, so a fresh subagent without implementation context should walk the page cold. It should return pass/fail per goal, findings with severity, and screenshot paths, not page text or click-by-click narration.

What usability issues block a UI from shipping?▼

Critical issues block the ship: failures to complete a user goal, missing feedback after actions, broken mobile layouts, or destructive actions without confirmation. These must be fixed and the review re-run before merging.