Impeccable injects seven-dimensional design knowledge into your AI workflow with a single setup command. Running `/impeccable teach` once loads a project-wide context that every subsequent command taps into. You get 18 precise commands—things like `/layout`, `/polish`, `/extract`—each addressing a discrete design discipline. Together they form a shared vocabulary between you and any supported AI tool (Claude Code, Codex CLI, Gemini, Cursor, Pi, Trae, etc.), so your prompts drive consistent, on-brand results instead of generic templates. On the detection side, Impeccable spots 25 common anti-patterns—nested cards, purple gradients, low-contrast text, oversize icons—without an LLM. A CLI (`npm i -g impeccable` or `npx impeccable detect src/`) audits HTML, CSS, JSX/TSX, Vue and Svelte files or even live URLs via Puppeteer, producing CI-ready JSON output. The Chrome DevTools extension does the same live on any page: click the toolbar icon to highlight problems in place. That makes peer reviews, competitor audits, or random browsing instantly more rigorous. The latest release pruned overlap—shrinking from 21 to 18 commands by merging `/normalize` into `/polish`, `/arrange` into `/layout`, and folding `/onboard` into `/harden`. The core skill got renamed from `frontend-design` to `impeccable` and now auto-cleans deprecated files. Behind the scenes, an internal eval framework tested frontier models with and without the skill, measuring font and color diversity gains. The anti-pattern engine now handles OKLCH, gradient-backed text, CSS variable borders and more. New capabilities include `/shape` for discovery interviews, a revamped docs site with live demos, and persona-driven critiques scoring against Nielsen heuristics.