<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI工程化 on RekCore</title><link>https://rekcore.com/tags/ai%E5%B7%A5%E7%A8%8B%E5%8C%96/</link><description>Recent content in AI工程化 on RekCore</description><generator>Hugo -- gohugo.io</generator><language>zh-cn</language><copyright>© 2026 RekCore</copyright><lastBuildDate>Wed, 06 May 2026 12:00:00 +0800</lastBuildDate><atom:link href="https://rekcore.com/tags/ai%E5%B7%A5%E7%A8%8B%E5%8C%96/index.xml" rel="self" type="application/rss+xml"/><item><title>AI Agent 动不动就翻车？六层 Harness 架构让你告别玄学调参</title><link>https://rekcore.com/tutorials/2026-05-06-harness-engineering-six-layers/</link><pubDate>Wed, 06 May 2026 12:00:00 +0800</pubDate><guid>https://rekcore.com/tutorials/2026-05-06-harness-engineering-six-layers/</guid><description>AI Agent 频繁跑偏、死循环、成功率卡在 70%？北大博士后卢菁和一线团队的经验表明，真正的瓶颈是 Harness——模型外面的运行系统。本文详解六层架构设计。</description></item><item><title>别再死磕 Prompt 了：2026 年 AI 工程化的真正战场在哪？</title><link>https://rekcore.com/tutorials/2026-05-06-beyond-prompt-engineering/</link><pubDate>Wed, 06 May 2026 11:00:00 +0800</pubDate><guid>https://rekcore.com/tutorials/2026-05-06-beyond-prompt-engineering/</guid><description>很多团队做 AI Agent 效果不好就疯狂改 Prompt，但问题往往出在上下文和框架层。本文拆解 AI 工程化的三个阶段，帮你定位 Agent 拉胯的真正原因。</description></item></channel></rss>