This gives Claude distinct operating modes that change how it approaches different types of work. Instead of treating everything like a tutorial, it can brainstorm with multiple options and pros/cons, implement code cleanly without verbose explanations, debug systematically by forming hypotheses, review code with severity ratings, teach concepts from fundamentals, or focus on production readiness for shipping. The mode detection is based on trigger words like "ideas" for brainstorm or "bug" for debug, though you can force a mode manually. The implement mode specifically references clean-code standards and emphasizes reading references before coding, which should help with the common problem of AI rushing to write code before understanding context.
npx -y skills add davila7/claude-code-templates --skill behavioral-modes --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills