This one takes your code, documentation, or architecture and systematically tries to break it by assuming the worst case scenarios and poking holes in your assumptions. You'd use it before shipping something important or when you need a reality check on whether your design actually holds up under stress. It's essentially a structured skeptic that asks the uncomfortable questions you might be avoiding, like edge cases you haven't considered or failure modes you've handwaved away. The approach is less about being mean and more about being thorough in ways that save you from production fires later.
npx -y skills add alirezarezvani/claude-skills --skill adversarial-reviewer --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