Cross Critique

When you've got multiple subagents who've independently proposed different solutions to the same problem, this runs a second round where each one critiques the others' proposals with structured pros and cons. The authors have deeper context than you do from their investigation, so letting them poke holes in each other's reasoning surfaces tradeoffs and failure modes you'd miss if you just synthesized their reports yourself. Use it for contested decisions like architecture tradeoffs, code review disagreements, or competing root cause theories. It adds latency and tokens, so skip it when proposals already agree or when there's an objectively verifiable answer. Pairs naturally with council and research skills whenever independent work diverges.

Install to Claude Code

npx -y skills add warpdotdev/common-skills --skill cross-critique --agent claude-code