memU

memU is a workspace runtime for AI agents. It compiles raw data such as chat logs, documents, code, images, audio, and tool logs into three persistent layers: index, skills, and memory. This enables agents to perform tasks based on context, continuity, and control. The index layer provides a map of all known information, the skills layer records learned skills and tool patterns, and the memory layer stores user preferences, goals, and key events. Through `memorize()` and `retrieve()` operations, agents can efficiently process and retrieve relevant information.