Most agent memory systems today are “store → retrieve” with vector search, but they don’t handle things like stale memory, conflicting facts, or explaining why something was retrieved.
This project adds conflict detection, lifecycle (decay + archival), and explainable retrieval on top of a standard embedding-based memory system.