image-analysis
Installation
SKILL.md
Image Analysis
Overview
An image holds two kinds of information. Some of it can be measured from the pixels: dimensions, format, transparency, which colors cover how much area, where an element sits and how large it is. The rest has to be seen: how the layout is organized, what the hierarchy is, what the picture feels like. This skill gives the agent a script for the first kind, a way of looking for the second, and one rule that keeps them apart: every number in the answer comes from the pixels, never from an impression.
Instructions
1. Find out what the result is for
- Building a UI from a screenshot or mockup: exact colors per role, element sizes and spacing, delivered as CSS variables or a Tailwind theme.
- A palette from a photo or artwork: five to eight representative colors with their weight, plus the small accents.
- Checking an asset: format, dimensions, transparency, orientation, color profile, animation.
- Describing a reference: a written brief of layout and style backed by measured values.
Ask where the result goes (stylesheet, token file, design note) when the request does not say.