<?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>Autonomous Loop on Chuanxilu for Skilled Homo sapiens</title><link>https://blog.chuanxilu.net/en/tags/autonomous-loop/</link><description>Recent content in Autonomous Loop on Chuanxilu for Skilled Homo sapiens</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Wed, 24 Jun 2026 09:00:00 +0800</lastBuildDate><atom:link href="https://blog.chuanxilu.net/en/tags/autonomous-loop/index.xml" rel="self" type="application/rss+xml"/><item><title>The Half-Life of Protocol Compliance (Part 2): Deep Root Causes</title><link>https://blog.chuanxilu.net/en/posts/2026/06/half-life-of-protocol-compliance-2/</link><pubDate>Wed, 24 Jun 2026 09:00:00 +0800</pubDate><guid>https://blog.chuanxilu.net/en/posts/2026/06/half-life-of-protocol-compliance-2/</guid><description>Why does protocol compliance degrade over rounds? From softmax attention dilution and RoPE distance decay to the EOS bias and transformer statelessness, layer by layer to the algorithmic root causes of LLM loop failure.</description></item><item><title>The Half-Life of Protocol Compliance (Part 1): Why Agents Won't Loop</title><link>https://blog.chuanxilu.net/en/posts/2026/06/half-life-of-protocol-compliance-1/</link><pubDate>Fri, 19 Jun 2026 08:00:00 +0800</pubDate><guid>https://blog.chuanxilu.net/en/posts/2026/06/half-life-of-protocol-compliance-1/</guid><description>AI agents won&amp;#39;t loop autonomously. Push once, they turn once. Stop pushing, they stop and ask &amp;#39;should I continue?&amp;#39; 20 hours, 161 confirmation requests, and a deeper problem hidden beneath clean formatting.</description></item></channel></rss>