📄 中文摘要
该研究提出了一种新的三层记忆架构，旨在提升大型语言模型（LLMs）的智能化水平。与传统的优化方法不同，这种架构强调AI的记忆能力，认为AI不仅需要掌握命令和知识，还需具备对用户及其项目的深刻理解。当前的AI在处理信息时缺乏持久的记忆，无法记住用户的具体操作和错误，导致知识无法转化为经验。Memento架构通过结合Redis、Postgres和向量存储，旨在为AI提供更丰富的上下文和记忆能力，从而改善用户体验和AI的实用性。
适用于大型语言模型的三层记忆架构（Redis + Postgres + 向量）MCP
📄 English Summary
A Three-Layer Memory Architecture for LLMs (Redis + Postgres + Vector) MCP
A new three-layer memory architecture is proposed to enhance the intelligence of large language models (LLMs). Unlike traditional optimization methods, this architecture emphasizes the importance of memory in AI, arguing that it needs not only to master commands and knowledge but also to have a deep understanding of users and their projects. Current AIs lack persistent memory, failing to remember specific user actions and mistakes, which prevents knowledge from being transformed into experience. The Memento architecture combines Redis, Postgres, and vector storage to provide richer context and memory capabilities for AI, aiming to improve user experience and the practicality of AI.