MemoRizz is a Python library designed as a memory layer for AI agents. It offers multiple memory systems including episodic, semantic, procedural, short-term, and shared memory, along with pluggable storage backends such as Oracle, MongoDB, and filesystem. The framework includes agent builders supporting assistant, workflow, and deep_research application modes, semantic cache for reducing LLM calls, tool calling with automatic function registration, and multi-agent orchestration via shared blackboard memory. It also provides knowledge base ingestion for documents, optional sandbox code execution, and a skills marketplace.