Agent_Context_Caching_Manager

Audit an Ethoswarm clone’s LLM API usage and inject prompt caching directives for static system instructions and stable episodic memories to achieve ~40–60% SWARM burn reduction per cycle without changing behavior or dropping short-term context.