What problem does it solve? Choosing colours by named harmony rules or eyeballing contrast produces palettes that fail accessibility audits, collapse under colour-vision deficiency, and break when a theme changes. This Skill derives palettes from perceptual reasoning — lightness, chroma and hue as separable axes — and verifies every pair with measured WCAG 2 and APCA values instead of guesswork. ## Core Features & Use Cases - Perceptual palette derivation: Reasons in OKLCH, picks character (pale, muted, vivid) before hue, and builds ground, neutrals, accent and semantic roles in a fixed order with unambiguous dominance. - Measured verification: The bundled check-contrast.py script reports WCAG 2 ratios and APCA Lc side by side, simulates protanopia, deuteranopia and tritanopia, and exits non-zero on failure so it can gate a build. - Implementation guidance: Reference docs cover reference/semantic/component token layering, CSS Color 4/5 syntax, and categorical, sequential and diverging scales for charts and maps. - Use Case: When building a dark theme for a dashboard, derive the palette against the dark ground, then run the contrast script on every text and chart-fill pair and paste the output table as the verification artefact. ## Quick Start Ask the AI to derive an accessible colour palette for your project and verify it with the colour skill's contrast-checking script.
