This is meta-tooling for building other Claude skills, and it's opinionated in useful ways. It enforces progressive disclosure (keep SKILL.md under 500 lines, push details to references), bakes in activation economics (your description must include trigger phrases users actually say), and calibrates freedom levels based on consequence (creative work gets principles, fragile operations get exact scripts). The workflow is structured around concrete examples first, then planning reusable components. The NEVER list is half the value here: don't write tutorials, don't explain standard operations, don't skip anti-patterns. If you're building custom skills or find yourself rewriting the same expertise repeatedly, this codifies what experienced skill authors learned through trial and error.
npx -y skills add gohypergiant/agent-skills --skill accelint-skill-manager --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
moizibnyousaf/ai-agent-skills
github/awesome-copilot