What problem does it solve? AI-generated visual output often falls into generic, low-quality defaults like purple gradient glows, identical card columns, and poor contrast. This Skill enforces deliberate design decisions, calibrated styling parameters, and strict anti-slop rules so every UI, document, or image prompt meets professional visual standards. ## Core Features & Use Cases - Brief Inference & Calibration: Declares a one-line design read and tunes three dials (variance, motion, density) before generating any CSS, markup, or image prompt. - Pure Vanilla CSS Architecture: Provides token-based :root custom properties, double-bezel cards, island buttons, bento grids, and floating navigation with zero third-party CSS dependencies. - Multi-Modal Image Pipeline: Directs the native generate_image tool with structured prompt formulas for web sections (16:9), mobile screens (9:16), and brand boards (1:1), then deconstructs results into production CSS. - Use Case: Ask the agent to redesign a dated landing page. It scans the existing code, diagnoses typography and color defects against the audit matrix, and applies a surgical modern CSS uplift without breaking form fields or JavaScript bindings. ## Quick Start Use the design-taste skill to build a dark-themed SaaS landing page hero section with a single accent color and pure Vanilla CSS.