Agentic Prompt Compaction Strategies

What are your favorite reasoning/compaction strategies for saving token spend, and why?