What problem does it solve? Engineering plans often contain hidden architecture flaws, missing test coverage, and performance risks that only surface during implementation. This Skill locks in an execution plan by walking through issues interactively with opinionated recommendations before any code is written. ## Core Features & Use Cases - Four-section review: Evaluates architecture, code quality, test coverage, and performance, raising each finding as an individual decision with options and recommendations. - Test coverage mapping: Traces every planned codepath and user flow into an ASCII coverage diagram, then adds missing tests (unit, E2E, or LLM eval) to the plan. - Outside voice challenge: Runs an independent second opinion from another AI model (Codex CLI or a subagent) and presents cross-model disagreements for user decision. - Use Case: Before implementing a new billing service, run this Skill on the design doc to catch a missing idempotency plan, untested error paths, and an N+1 query risk, then lock the corrected plan. ## Quick Start Ask the AI to review the architecture and lock in the implementation plan for your current design doc.
