Impeccable

Impeccable, by Paul Bakaus (creator of jQuery UI), tackles a problem every AI-assisted frontend team recognizes: models trained on the same SaaS templates produce the same design tells — Inter everywhere, purple-to-blue gradients, cards nested in cards, gray text on colored backgrounds. It started from Anthropic's frontend-design skill and grew into a full design harness for AI coding agents. Installation is one command (npx impeccable install), which detects your tools and installs the skill into Claude Code, Cursor, GitHub Copilot, Gemini CLI, Codex CLI, Grok Build, OpenCode, Trae, Rovo Dev, Qoder, and more. Everything runs through /impeccable with 23 commands: init writes PRODUCT.md and DESIGN.md so later commands know your audience and brand; craft, shape, critique, audit, polish, bolder, quieter, distill, animate, harden, and live (visual variant iteration in the browser) cover the full design workflow. Beyond the LLM-driven commands, a standalone CLI and browser extension run 46 deterministic detector rules — no LLM or API key needed — catching AI-slop patterns and general design-quality issues (line length, cramped padding, small touch targets, skipped headings), with CI-friendly JSON output. Provider-native hooks run the detector automatically on UI file edits; Cursor can even block bad writes before they land. With 45k+ GitHub stars, it is Apache 2.0 licensed and completely free. It integrates via agent skills and hooks rather than MCP — no MCP server or client is advertised.