What problem does it solve? Consequential plans, designs, and decisions often rest on hidden assumptions that only surface after implementation begins. This Skill interrogates unresolved decisions through structured rounds of questions, separating facts from assumptions so risks and tradeoffs become visible before action. ## Core Features & Use Cases - Decision-Tree Interviewing: Builds a decision tree from your stated objective and asks only questions whose prerequisites are already resolved, recomputing the frontier after each round. - Evidence-First Resolution: Resolves factual questions from repository state, files, tools, or accepted records instead of making you answer what the environment can answer, while keeping owner decisions with the owner. - Bounded Recommendations: Provides a concise recommended answer and material tradeoff per question when evidence supports one, then stops before execution. - Use Case: Before committing to a new system architecture, run a grilling session to expose contradictions with existing repository state, label unresolvable assumptions as UNKNOWN, and confirm shared understanding before routing to planning or design work. ## Quick Start Grill my proposed migration plan for moving the billing service to a new database before I start implementation.