meta-prompting

Recursive prompt improvement with quality-driven iteration for Claude Code. Use when you need systematic, measurable improvement of LLM outputs through complexity routing, context extraction, and iterative refinement.