<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>原子级 on Tony Bai</title><link>https://tonybai.com/tags/%E5%8E%9F%E5%AD%90%E7%BA%A7/</link><description>Recent content in 原子级 on Tony Bai</description><generator>Hugo</generator><language>zh-cn</language><copyright>2004-2026 Tony Bai. 版权所有.</copyright><lastBuildDate>Tue, 13 Jan 2026 00:00:00 +0800</lastBuildDate><atom:link href="https://tonybai.com/tags/%E5%8E%9F%E5%AD%90%E7%BA%A7/index.xml" rel="self" type="application/rss+xml"/><item><title>像构建 Claude Code 一样构建应用：揭秘 Agent-native 架构的 5 大核心原则</title><link>https://tonybai.com/2026/01/13/agent-native-architecture/</link><pubDate>Tue, 13 Jan 2026 00:00:00 +0800</pubDate><guid>https://tonybai.com/2026/01/13/agent-native-architecture/</guid><description>本文永久链接 – https://tonybai.com/2026/01/13/agent-native-architecture 大家好，我是Tony Bai。 软件智能体（Software Agents）现在已经能够可靠地工作了。**Claude Code** 证明了，只要赋予一个大语言模型（LLM）访问 Bash 和文件系统的权限，并让它在一个循环中运行直到达成目标，它就能自主完成复杂的多步...</description></item></channel></rss>