memorizz

Why we included this project

Memorizz is built for agents that need to remember across sessions. It's a Python library that handles the persistence for you, so you don't have to wire up a database and embedding pipeline by hand. The library bundles five memory systems (episodic, semantic, procedural, short-term, and shared) and lets you plug in MongoDB, Oracle, or a plain filesystem as the backend, which makes it easy to fit into a stack you already run. Beyond the basic storage, you get semantic retrieval with embeddings and vector search, entity memory for profile-style facts, and a semantic cache that cuts down on repeat LLM calls. It also connects over MCP and ships a memory-first harness that can run Codex, Claude Code, OpenHands, or a native agent behind shared memory and policy controls. The project is explicitly experimental and released under a noncommercial license, so it's best suited to research, learning, and evaluation rather than production workloads.