kaelig/state-of-ai-in-design-systems

A field survey of how 20 open-source design systems make themselves AI-friendly: MCP servers, agent skills, llms.txt, and the techniques that keep models on-system (last updated: July 2026)

10 files for Claude Code, Codex and OpenCode: audit-prose, correct, draft, research and 6 more — 5,437 tokens loaded in every session.

Open on GitHub state-of-ai-in-design-systems.netlify.app

Instructions

state-of-ai-in-design-systems AGENTS.md AGENTS.md A 5,398 tok
state-of-ai-in-design-systems CLAUDE.md CLAUDE.md A 39 tok

Skills

contribute-system .claude/skills/contribute-system/SKILL.md A 89 tok
impeccable .claude/skills/impeccable/SKILL.md A 189 tok

Commands

audit-prose .claude/commands/system/audit-prose.md A 20 tok
correct .claude/commands/system/correct.md A 25 tok
draft .claude/commands/system/draft.md A 26 tok
research .claude/commands/system/research.md A 17 tok
ship .claude/commands/system/ship.md A 27 tok
verify .claude/commands/system/verify.md A 16 tok

These files are kaelig/state-of-ai-in-design-systems's own configuration — they tell Claude Code, Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.