MemMachine

Why we included this project

Most agent frameworks treat every session as a clean slate, so developers who want personalization or continuity end up building durable state by hand. MemMachine fills that gap with a dedicated memory layer: it stores conversational history, long-term user facts and preferences, and short-term session context, letting assistants pick up where they left off after restarts and even model changes. The design stays deliberately light. A small Python client talks to a server you can run locally or host yourself, and memory is exposed through typed interfaces instead of ad-hoc SQL scattered across your code. Teams building chatbots, support assistants, or any autonomous agent that has to remember its users get a practical middle ground between a full knowledge base and an ephemeral chat buffer, with a graph-based episodic store and SQL-backed profiles underneath.