Agent Prompt Evolution

Track agent specialization by recording when generic agents show performance gaps (>5x), documenting additions and removals in your agent set across iterations, and monitoring capability changes to your coordinating meta-agent. This identifies which specialized agents deliver ROI versus which remain universal components.

Introduce specialized agents when comparing multi-experiment results reveals performance disparities that generic agents cannot resolve, or when analyzing methodology transferability shows domain-specific logic emerging. Use this Skill to decide which specializations justify the 2–3 hour overhead per experiment.

Reusability assessment compares agent behavior and performance across experiments to isolate universal logic (applicable across domains) from domain-specific reasoning. This evaluation informs which components to maintain as core versus which to retire or specialize.

Reusability depends on tracking meta-agent evolution and specialization decisions alongside agent set changes. This Skill provides ROI-focused reporting to show whether agents designed for one methodology transfer to others or require domain-specific adaptation.

Expect 2–3 hours per experiment to record specialization decisions, track agent set and meta-agent changes, and generate templates and quick-start guidance. This overhead is amortized across iterations when comparing multiple experiments.

Document the specialization decision, measure performance before and after across subsequent iterations, and use ROI-focused reporting to compare results. This shows whether the specialized agent's performance gain justifies maintaining it versus reverting to generic design.