Skill Architect

If you're building skills for Claude Code, this is your design system. It handles the full lifecycle: scaffolding new skills from scratch, auditing existing ones for activation problems, and encoding real domain expertise using progressive disclosure patterns. The three-layer architecture (lightweight metadata, lean SKILL.md under 500 lines, on-demand references) is the core philosophy here. Most useful when you're debugging why a skill won't activate, tightening descriptions to reduce false positives, or refactoring bloated skills that dump everything into context at once. Pairs naturally with skill-grader for feedback loops and skill-creator for implementation. Think of it as the meta-tool that teaches you to write skills that actually trigger when they should and stay quiet when they shouldn't.

skills.sh installs: 116
GitHub stars: 113
last updated: Jun 3, 2026

View on GitHub

Install

$ npx -y skills add erichowens/some_claude_skills --skill skill-architect --agent claude-code

Installs into .claude/skills of the current project.

Files

Skill Architect: The Authoritative Meta-Skill

The unified authority for creating expert-level Agent Skills. Encodes the knowledge that separates a skill that merely exists from one that activates precisely, teaches efficiently, and makes users productive immediately.

Philosophy

Great skills are progressive disclosure machines. They encode real domain expertise (shibboleths), not surface instructions. They follow a three-layer architecture: lightweight metadata for discovery, lean SKILL.md for core process, and reference files for deep dives loaded only on demand.