AIWF evaluation skill for local AI models and routes. Use for lm-eval readiness, custom prompt suites, image and video smoke receipts, before and after model comparisons, baseline selection, acceptance thresholds, regression gates, training promotion checks, and release-facing validation artifacts.
