What problem does it solve? In CI or autonomous agent workflows, no human reviewer is available to challenge a plan or design before implementation. This Skill fills that gap by having the agent interrogate its own artifact, exposing edge cases, contradictions, and unstated assumptions before code is written. ## Core Features & Use Cases - Self-Interrogation Loop: Generates at least 10 tough, skeptical reviewer questions targeting edge cases, undefined terms, contradictions, and scope leaks. - Honest Gap Flagging: Answers each question from the codebase and issue context, marking unanswerable items as needs-human instead of inventing answers. - Plan Hardening with Audit Trail: Applies every answer back into the artifact with before/after diffs, producing a hardened plan plus a full question-and-answer record. - Use Case: An autonomous CI agent drafts an implementation plan for a GitHub issue, then runs a self-grill to catch missing acceptance criteria and surface open decisions for human confirmation. ## Quick Start Ask the agent to grill the current implementation plan and list any questions it cannot answer without human input.
