`kaeru` is a cutting-edge cognitive engine designed for LLM agents, featuring a unique two-tier graph structure that enhances multi-agent collaboration. This platform supports both local-first operation and optional cloud sharing, allowing teams to build upon each other's knowledge seamlessly. The cognitive tier facilitates active thought processes, while the recollection tier archives settled ideas, ensuring a rich context for ongoing projects.

Key features include bi-temporal data management, initiative-specific subgraphs, and a comprehensive curator API with around 40 primitives for managing knowledge and collaboration. With its focus on operational semantics and structural recall, `kaeru` promotes efficient knowledge sharing and retrieval, making it an essential tool for teams working with AI agents.