<?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>Opencode on Square Uncle</title>
    <link>https://squareuncle.com/en/tags/opencode/</link>
    <description>Recent content in Opencode on Square Uncle</description>
    <generator>Hugo -- 0.147.7</generator>
    <language>en</language>
    <lastBuildDate>Sat, 15 Aug 2026 14:00:00 +0800</lastBuildDate>
    <atom:link href="https://squareuncle.com/en/tags/opencode/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>The First Place to Cut Token Spend: Hand the Grunt Work to a Cheap Model (flash-delegate, Field Notes)</title>
      <link>https://squareuncle.com/en/posts/2026-08-15-flash-delegate-token-saving/</link>
      <pubDate>Sat, 15 Aug 2026 14:00:00 +0800</pubDate>
      <guid>https://squareuncle.com/en/posts/2026-08-15-flash-delegate-token-saving/</guid>
      <description>The orchestrator (expensive model) only owns task breakdown and review; mechanical work goes to a cheap fast model. One real closed loop, three real pitfalls, and what can and can&amp;rsquo;t be delegated.</description>
    </item>
  </channel>
</rss>
