I made an open-source memory layer for agents

This open-source memory layer gives AI agents persistent, searchable memory by automatically extracting entities and relationships from interactions, enabling semantic recall via an MCP server, REST API, and SDKs. It’s built for developers building autonomous agents who need a self-hostable or cloud-based memory backend, and it’s interesting because it bridges the gap between stateless LLM calls and long-term, structured agent cognition under a permissive MIT license.