<?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>RAG on 茶桁.MAMT</title><link>https://hivan.me/tags/rag/</link><description>Recent content in RAG on 茶桁.MAMT</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 24 Jul 2026 14:00:00 +0800</lastBuildDate><atom:link href="https://hivan.me/tags/rag/index.xml" rel="self" type="application/rss+xml"/><item><title>RAG vs MCP：驱动现代 AI 系统的隐秘之战</title><link>https://hivan.me/posts/rag-vs-mcp%E9%A9%B1%E5%8A%A8%E7%8E%B0%E4%BB%A3-ai-%E7%B3%BB%E7%BB%9F%E7%9A%84%E9%9A%90%E7%A7%98%E4%B9%8B%E6%88%98/</link><pubDate>Fri, 24 Jul 2026 14:00:00 +0800</pubDate><guid>https://hivan.me/posts/rag-vs-mcp%E9%A9%B1%E5%8A%A8%E7%8E%B0%E4%BB%A3-ai-%E7%B3%BB%E7%BB%9F%E7%9A%84%E9%9A%90%E7%A7%98%E4%B9%8B%E6%88%98/</guid><description>大多数开发者认为构建 AI 系统的核心在于选择模型，但真正的挑战在于理解何时让 AI 获取知识（RAG），何时让 AI 执行操作（MCP）。本文深入剖析两者的区别、互补性以及混合架构的未来。</description></item></channel></rss>