siuserxiaowei

siuserxiaowei / visual-taste-lab

Public

VI-first UI/UX design-language skill for Codex and Claude Code｜让 AI 先做视觉身份审计再写页面

siuserxiaowei.github.iovisual-taste-lab ai-web-design claude-code codex design-system ui-ux
13
1
100% credibility
Found May 07, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
HTML
AI Summary

Visual Taste Lab is a skill pack for AI coding tools that teaches them to analyze a project's visual identity and generate reusable design guidelines before creating or refining webpages.

How It Works

1
🔍 Discover Visual Taste Lab

You stumble upon this helpful guide while looking for ways to make AI-built webpages look more stylish and brand-specific.

2
📥 Get the Skill Pack

You download the simple skill folder from the page to add better design smarts to your AI helper.

3
🧰 Add to Your AI Toolbox

You place the skill pack into your AI assistant's special folder so it can use these design tricks anytime.

4
💬 Ask AI to Improve Your Page

You tell your AI helper to use Visual Taste Lab first—it pauses to study logos, colors, and page type before making changes.

5
🎨 AI Builds Design Rules

Your AI analyzes the brand's look, picks the right style for the page, and creates easy rules for colors, fonts, and layouts.

✨ Enjoy Your Polished Webpage

Your webpage now has a professional, trustworthy visual style that matches the brand perfectly, feeling fresh and reliable.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is visual-taste-lab?

Visual Taste Lab is a VI-first UI/UX design-language skill built for Codex and Claude coding agents, written in HTML. It forces AI to audit a project's visual identity—logo cues, colors, page type—before generating pages, distilling reusable rules for typography, spacing, buttons, and more. Developers get structured prompts and workflows to turn vague "make it pretty" into consistent, brand-aligned HTML output.

Why is it gaining traction?

Unlike generic templates or one-liner prompts, it enforces a deliberate visual audit first, classifying sites as dashboards, landing pages, or portfolios to avoid template-y results. Users notice tighter AI outputs: honest claims, mobile-safe layouts, and system-wide tokens applied before tweaks. The hook is copy-paste goal prompts and a public preview for instant testing in Codex or Claude chats.

Who should use this?

Frontend devs prompting AI to polish HTML prototypes for client sites or SaaS dashboards. Indie hackers refining product landing pages with brand refs. Teams using Claude or Codex for rapid UI iteration, especially when fighting generic AI visuals.

Verdict

At 13 stars and 1.0% credibility, it's raw and unproven—docs are solid with examples, but expect tweaks for production. Grab it if you're deep in AI UI workflows; skip for now if you need battle-tested tools.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.