Memory management remains a fundamental challenge for autonomous AI systems. Existing approaches employ operating system metaphors (MemGPT), flat storage with LLM-driven management (Mem0), or note-taking paradigms (A-MEM), but none incorporate principles from cognitive neuroscience—despite memory science offering over a century of empirically validated models.We present ZenBrain, a multi-layer memory architecture for AI agents that integrates twelve established neuroscience models into a unified system. ZenBrain implements seven distinct memory layers—working, short-term, episodic, semantic, procedural, core, and cross-context—orchestrated by twelve algorithms including Hebbian learning dynamics for knowledge graph co-activation (Hebb, 1949), Ebbinghaus forgetting curves with FSRS spaced repetition scheduling (Ebbinghaus, 1885), sleep-time memory consolidation in three phases (SWS/REM/SHY, Stickgold & Walker, 2013), Bayesian confidence propagation with 95% confidence intervals, and emotional valence tagging (McGaugh, 2004).ZenBrain is open-source, production-deployed in ZenAI, and distributed as composable npm packages (@zensation/algorithms, @zensation/core) with 9,500+ automated tests and zero production dependencies. See the latest version for full benchmark results (LoCoMo, MemoryAgentBench, MemoryArena).