This is Gary Klein's premortem technique adapted for code review and planning. You run it before implementation to surface failure modes you'd normally only spot after things break. It forces you through a structured checklist of technical risks (scalability, dependencies, security) and then categorizes findings as tigers (real threats), paper tigers (look scary but fine), or elephants (things nobody wants to discuss). The verification requirements are strict: you can't flag a tiger without proving you checked for existing mitigations in the surrounding code. Works in quick mode for PRs or deep mode for pre-implementation reviews. It blocks with an interactive prompt so you have to either accept the risks, add mitigations to your plan, or research solutions before moving forward.
npx skills add https://github.com/parcadei/continuous-claude-v3 --skill premortem