What problem does it solve? Vague or ambiguous requirements cause AI pipelines to build the wrong thing. This Skill runs a structured pre-pipeline interview that forces ambiguities into the open before any design or code is written, ending with an aligned brief the downstream pipeline can consume. ## Core Features & Use Cases - One-question-at-a-time interview: Asks exactly one open question per turn, each with a recommended answer the user can accept, override, or refine. - Codebase self-answering: Explores the repository to resolve questions the code already answers (conventions, prior art, file locations) instead of asking the user. - Aligned brief output: Writes the agreed decisions and residual open items to docs/features/<slug>/INPUT.md for /harness, /harness-plan, or /harness-stream to consume, with overwrite confirmation on collisions. - Use Case: You have a fuzzy feature idea and suspect the agent would build something subtly off. Run the grill session, answer the questions, and hand the resulting brief to the full pipeline. ## Quick Start Ask the agent to grill you on your feature idea before running the pipeline, then answer each question one at a time until the aligned brief is written.