This is a dialectical learning framework that treats every hypothesis as incomplete until practice validates it. When you've proposed a solution or formed a theory, it forces a cycle: gather hands-on data, extract patterns, test predictions, evaluate results against reality, then spiral upward. It's explicitly against two failure modes: dogmatism (copying solutions without verifying they fit) and empiricism (acting on feel without extracting principles). The method comes from Mao's 1937 epistemology essays, so expect Chinese philosophical vocabulary around practice as the sole criterion of truth. Use it when you're building understanding through iteration, not when you already know the answer or just need to execute once. It will make you declare which phase you're in before each step.
npx -y skills add hughyau/qiushi-skill --skill practice-cognition --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