Break down supposedly impossible problems by separating physics from convention, then rebuild solutions from scratch. This follows Aristotle's method, popularized by Musk at SpaceX when he questioned why rockets cost $65M and found the raw materials were under $100K. You'd use it when conventional solutions aren't working, when analyzing costs for dramatic savings, or when everyone says something can't be done. Claude structures the breakdown into assumptions, atomic components, and validation steps, but you need the domain expertise to actually challenge what's真 versus what's just industry habit. Works best when you're willing to question everything, not just optimize around the edges.
npx -y skills add guia-matthieu/clawfu-skills --skill first-principles --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