<?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>Engineering Practice on AI Pulse</title>
    <link>https://cuigh.com/tags/engineering-practice/</link>
    <description>Recent content in Engineering Practice on AI Pulse</description>
    <generator>Hugo</generator>
    <language>en-US</language>
    <lastBuildDate>Fri, 03 Jul 2026 16:05:13 +0800</lastBuildDate>
    <atom:link href="https://cuigh.com/tags/engineering-practice/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Agent Engineering Playbook: 5 Teams Hit the Same Wall in Early July</title>
      <link>https://cuigh.com/posts/agent-engineering-playbook-2026/</link>
      <pubDate>Fri, 03 Jul 2026 16:05:13 +0800</pubDate>
      <guid>https://cuigh.com/posts/agent-engineering-playbook-2026/</guid>
      <description>Five engineering tips I read on July 2 and 3 — from Hacker News, X, blog posts, and WeChat — all pointed at the same moment: AI agents crossed into the engineering phase. From okturtles&amp;#39; Short Leash method to Qwen Zhu Da&amp;#39;s Harness engineering, from SGLang&amp;#39;s Loop Engineering to browser-use&amp;#39;s video-use Skill, plus a WeChat-transcribed run where Fable 5 autonomously completed real ops engineering for an overseas site, five teams spoke different languages and landed the same conclusion. Running agents is easy. Keeping them in line is the actual work.</description>
    </item>
  </channel>
</rss>
