skill-creation-optimizing-descriptions teaches systematic evaluation and iteration of skill descriptions to achieve reliable triggering on relevant agent prompts, using train/validation splits and repeated testing cycles.

skill-creation-best-practices teaches how to ground skills in real expertise, refine through execution, calibrate specificity to task fragility, and apply patterns like gotchas sections and validation loops. it covers scope, context efficiency, and instruction design but lacks explicit decision points and outcome signals.