skill-pressure-test

Have the plan interrogated one question at a time, with each question carrying a recommended answer you can confirm or dispute. The process follows decision dependencies, surfaces contradictions, and ends with a verdict of sound, sound-with-changes, or unsound.

Brainstorming opens a space up and generates options; pressure-testing closes one down and looks for what is wrong with an existing position. Use pressure-testing when a plan already exists and the risk is an unspoken flaw, not a lack of ideas.

Skip it when no position exists yet, when you want options generated, or when the answer is discoverable by reading the codebase. It also stops producing value once remaining questions no longer change what anyone would do.

No. The skill produces agreement, not changes — nothing in the plan is modified during the interview, and no action is taken until you confirm shared understanding. Changes are only recorded in the final output summary.

It produces five sections: where the plan stands, decisions settled in order, changes the answers imply, questions still open with what they block, and facts found with their sources cited.