gpt-taste

Generates award-level frontend UI code with GSAP animations, bento grids, and editorial typography.

Updated Jul 7, 2026
One-click install
npx skills add https://github.com/singhaganesh/Urban-assist --skill gpt-taste-singhaganesh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gpt-taste
Source: https://github.com/singhaganesh/Urban-assist/tree/main/.cursor/skills/gpt-taste
Command: npx skills add https://github.com/singhaganesh/Urban-assist --skill gpt-taste-singhaganesh

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? LLM-generated frontend code tends to produce repetitive layouts, narrow containers that wrap headings into 6-line walls, gappy bento grids, invisible button text, and static pages without motion. This Skill enforces a rigorous design engineering process that eliminates these default failure patterns. ## Core Features & Use Cases - Deterministic Layout Randomization: Simulates Python-based seeded randomization to select hero architectures, typography stacks (Satoshi, Cabinet Grotesk, Outfit, Geist), and component patterns, preventing repeated layouts. - Strict Structural Rules: Enforces AIDA page structure, 2-3 line H1 limits via wide containers, gapless bento grids with grid-flow-dense, and bans cheap meta-labels like "SECTION 01". - Advanced GSAP Motion: Implements ScrollTrigger pinning, scrubbing text reveals, card stacking, and image scale/fade scroll effects with hover physics on interactive elements. - Use Case: Ask for a landing page for a SaaS product and receive a complete React/Tailwind page with a cinematic hero, dense bento feature grid, pinned scroll sections, and a pre-flight design plan verifying spacing and grid math. ## Quick Start Ask the AI to build a landing page for your product and it will output a verified design plan followed by the full animated UI code.

Frequently Asked Questions about gpt-taste

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

FAQPage Schema
How do I stop AI-generated landing pages from looking repetitive?

Use deterministic randomization to select hero layouts, font stacks, and component patterns before writing code. This Skill simulates a seeded Python random.choice() based on prompt character count, forcing varied layouts instead of defaulting to the same left-right structure.

How to build a bento grid with no empty cells in Tailwind?

Apply grid-flow-dense (grid-auto-flow: dense) to the grid container and mathematically verify that col-span and row-span values interlock to fill every cell. Limit the grid to 3-5 intentional cards mixing imagery, typography, and CSS effects.

What GSAP ScrollTrigger animations work for landing pages?

Effective patterns include pinning a section title while content scrolls beside it, scrubbing text opacity from 0.1 to 1.0 sequentially, scaling images from 0.8 to 1.0 on entry, and stacking cards that overlap as the user scrolls down.

Why do AI-generated hero sections wrap into 6 lines of text?

LLMs default to narrow containers that force headings to wrap excessively. Fix this by using ultra-wide containers like max-w-5xl or max-w-6xl and clamp-based font sizing such as clamp(3rem, 5vw, 5.5rem) to keep the H1 within 2-3 lines.

Does this approach work with fonts other than Inter?

Yes, Inter is explicitly banned. The Skill randomizes between Satoshi, Cabinet Grotesk, Outfit, and Geist to produce distinctive editorial typography for each generated page.