Your AI pair programmer gets a design degree

What it does

This repository is a structured prompt library dressed up as agentic plugins. It breaks the full design lifecycle—from user research and strategy to visual critique and developer handoff—into 91 “skills” (context files that teach an AI about a topic) and 28 “commands” (workflows that chain skills into tasks). Loaded into Claude Code or Gemini CLI, the collection gives the agent a grounded vocabulary for design work rather than letting it hallucinate personas and color theory on demand.

The interesting bit

The project treats design methodology as infrastructure. Instead of asking an AI to “make a nice UI,” you invoke workflows that reference accessibility checks, Gestalt principles, token governance, or heuristic evaluation criteria. It is essentially design ops translated into agent-readable manifests.

Key highlights

• Covers the complete design lifecycle: research, systems, strategy, UI, interaction, prototyping, ops, and critique
• Skills act as domain knowledge units (nouns), while commands chain them into actionable workflows (verbs)
• Supports Claude Code marketplace plugins and Gemini CLI workspace extensions
• Includes concrete workflows like competitive benchmarking, component scaffolding, A/B experiment design, and design debt reporting
• MIT licensed with contribution guidelines for expanding the knowledge base

Verdict

Worth a look if you want your AI assistant to follow structured design methodologies instead of generating generic advice. Pass if you need rendered components or Figma integrations—this outputs strategy, specs, and rationale, not pixels.