Zero-Mem: Eliminating LLM Token Costs for Agent Memory Operations

Zero-Mem implements zero-token memory operations for LLM agents by utilizing an entity-context graph to organize interaction traces. This method avoids using additional LLM calls for memory mediation, reducing recurring token consumption and latency during long-term interactions.