A standalone development judgment skill for AI agents. It uses the spirit of Plan-Do-Check-Act to keep development work aligned with the real problem, available evidence, confirmed boundaries, and verifiable outcomes. It is a judgment discipline, not a fixed workflow or process engine.
