概要
Agent Prompt Evolution is a meta-cognition skill designed to monitor and manage the emergence of specialized agents during complex methodology development. It provides a structured framework to determine when creating a new, specialized agent is worth the effort by using data-driven metrics like performance gaps (targeting >5x improvements) and Return on Investment (ROI). By categorizing agents as universal, domain-specific, or task-specific and tracking their stability across iterations, this skill helps developers build highly efficient, reusable agent libraries while ensuring the base Meta-Agent capabilities remain optimized for long-term project success.
