LLM Cognition Python MIT

memorg

memorg explores what happens when LLM agents need to remember across sessions. It implements structured, queryable memory systems that go beyond simple context windows, studying the difference between memory and context in agent architectures.

Technologies

Primary use case

Structured, queryable, schema-validated persistent memory for long-running LLM agents.

How it compares

memorg is one option in a category that includes Mem0, Letta (MemGPT), Zep , and vector memory layers. Our Compare page has the full side-by-side.