What problem does it solve? Running a full plan review normally means invoking three separate review skills and answering 15-30 intermediate questions. This Skill reads the CEO, design, and engineering review skills from disk and executes them sequentially at full depth, auto-deciding routine questions with 6 decision principles and surfacing only genuine taste decisions at a final approval gate. ## Core Features & Use Cases - Sequential Review Pipeline: Executes CEO (strategy/scope), design (UI/UX), and engineering (architecture/tests) review phases in strict order, each at the same depth as running the skills manually. - 6 Decision Principles: Auto-answers intermediate questions using completeness, boil-the-lake scope expansion, pragmatism, DRY, explicit-over-clever, and bias-toward-action rules, with every decision logged to an audit trail. - Taste Decision Surfacing: Classifies decisions as mechanical (auto-decided silently) or taste (close approaches, borderline scope, codex disagreements) and presents taste decisions at a final approval gate. - Restore Point & Dual Voices: Captures a restore point of the plan file before modifying it, and runs both a Claude subagent and Codex as independent outside voices with a consensus table. - Use Case: You have a rough implementation plan on a feature branch and want a fully reviewed plan without answering dozens of questions. Invoke the skill and receive a plan reviewed across strategy, design, and engineering with only the genuinely debatable decisions escalated to you. ## Quick Start Ask the AI to run autoplan on the current plan file to execute the full CEO, design, and engineering review pipeline with automatic decisions.
