What problem does it solve? Most plans only receive polite feedback, so critical assumptions go untested until reality breaks them. This Skill adversarially reviews a PRD, roadmap, or strategy to find the claims that would kill the plan if false, and tells you exactly what evidence to gather this week. ## Core Features & Use Cases - Assumption Extraction & Steelmanning: Lists every claim in the plan, separates load-bearing assertions from trivial ones, then attacks the strongest version of each rather than strawmen. - Ranked Failure Modes: Scores each failure mode by impact × likelihood × test cost, so the cheapest high-impact test surfaces first. - Actionable Output: For each surviving kill-assumption, returns the failure condition, evidence to gather this week, a kill criterion, and the cheapest test, plus explicit notes on what is well-reasoned and what could not be assessed. - Use Case: Before an executive review of a new product roadmap, run the plan through this Skill to get 3-5 ranked kill-assumptions with concrete validation experiments instead of a generic risk list. ## Quick Start Ask the AI to red-team your attached PRD or strategy document and return the top kill-assumptions ranked with cheapest tests and kill criteria.
