MemMA: Coordinating the Memory Cycle through Multi-Agent Reasoning and In-Situ Self-Evolution

ArXi:2603.18718v1 Announce Type: new Memory-augmented LLM agents maintain external memory banks to long-horizon interaction, yet most existing systems treat construction, retrieval, and utilization as isolated subroutines. This creates two coupled challenges: strategic blindness on the forward path of the memory cycle, where construction and retrieval are driven by local heuristics rather than explicit strategic reasoning, and sparse, delayed supervision on the backward path, where downstream failures rarely translate into direct repairs of the memory bank.