Context Compaction

The process of distilling large volumes of agentic reasoning traces and data into minimal, high-density representations. This reduces token costs and improves inference efficiency in multi-agent systems.