optimize_anything: A Universal API for Optimizing any Text Parameter

A single LLM-based optimizer treating all problems as text-artifact improvement achieves state-of-the-art across six domains: it triples Gemini Flash ARC-AGI accuracy from 32.5% to 89.5% and cuts cloud scheduling costs by 40%. Supports single-task, multi-task, and generalization modes. Challenges the assumption that domain-specific optimizers are necessary.