This skill should be used when the user asks to "implement LLM-as-judge", "compare model outputs", "create evaluation rubrics", "mitigate evaluation bias", or mentions direct scoring, pairwise compari
skills/guanyang/antigravity-skills/advanced-evaluation/SKILL.md(main)