<?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>Harness on Alex</title>
    <link>https://blog.igalaxy.xyz/tags/harness/</link>
    <description>Recent content in Harness on Alex</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Wed, 19 Aug 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://blog.igalaxy.xyz/tags/harness/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>让 Agent 自己干活,但我不失控:一套半自主工作循环的完整设计</title>
      <link>https://blog.igalaxy.xyz/posts/semi-autonomous-agent-loop/</link>
      <pubDate>Wed, 19 Aug 2026 00:00:00 +0000</pubDate>
      <guid>https://blog.igalaxy.xyz/posts/semi-autonomous-agent-loop/</guid>
      <description>我的时间被会议切碎,拿到任务没时间从零想,但决策权又不能让渡。这套循环解决的是这个矛盾:agent 承担理解成本,我只保留判断。文章给出完整设计——队列机制、定时调度、多 agent 规划、文件分层,以及三次被自己系统打脸的实证。</description>
    </item>
  </channel>
</rss>
