GEPA / reflective prompt mutation

eval harnesses and scoring pipelines

GEPA requires a scored execution loop to generate reflections and select Pareto-optimal prompts; the eval infrastructure is the enabling dependency.

1 source videos