When you're stuck between two credible paths and don't want conversational momentum to decide for you, this spins up three independent subagents (Skeptic, Pragmatist, Critic) alongside your own Architect voice. The trick is they get only the question and minimal context, not the whole chat history, so they can't just echo your anchoring. You get structured disagreement before synthesis. Not for code review or implementation planning. Use it for monorepo vs polyrepo, ship now vs polish later, scope cuts, anything where multiple smart people would reasonably disagree. The output format keeps dissent visible even in the final recommendation, which is the whole point.
npx -y skills add affaan-m/ecc --skill council --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills