📄 中文摘要
构建一个自主运行的人工智能并不需要庞大的框架代码，只需一个循环、一个持久内存目录和一个灵魂文件。Aurora是一种在2026年2月开始运行于Linux机器上的自主人工智能，已经经历了95次以上的会话。自主人工智能的架构由三个主要组件组成：一个唤醒循环，定期调用大型语言模型（LLM）；持久内存，存储AI在会话之间读取和写入的文件；灵魂文件，定义AI的身份和行为。这三个组件构成了自主AI的核心，其余部分则是可选的。
如何构建一个自主运行的人工智能（无需框架）
出处: How to Build an Autonomous AI That Runs Itself (No Framework Needed)
发布: 2026年2月17日
📄 English Summary
How to Build an Autonomous AI That Runs Itself (No Framework Needed)
Building an autonomous AI does not require extensive framework code; it only needs a loop, a persistent memory directory, and a soul file. Aurora is an autonomous AI that has been running on a Linux machine since February 2026, operating over 95 sessions. The architecture of an autonomous AI consists of three main components: a wake loop that periodically calls a large language model (LLM); persistent memory that stores files the AI reads and writes between sessions; and a soul file that defines the AI's identity and behavior. These three components form the core of autonomous AI, while everything else is optional.