Compare two LLM prompts side-by-side. See what changed, then let an AI judge score them on criteria you define.