<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Agent Cost Control on SoloSoft</title><link>https://www.solosoft.dev/tags/agent-cost-control/</link><description>Recent content in Agent Cost Control on SoloSoft</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 03 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://www.solosoft.dev/tags/agent-cost-control/index.xml" rel="self" type="application/rss+xml"/><item><title>Auditing the Caveman Skill: 109,102 Stars, a 65% Headline, and the 3% Its Own Eval Publishes</title><link>https://www.solosoft.dev/post/caveman-token-savings-audit/</link><pubDate>Sat, 03 Oct 2026 00:00:00 +0000</pubDate><guid>https://www.solosoft.dev/post/caveman-token-savings-audit/</guid><description>&lt;p&gt;&lt;code&gt;JuliusBrussee/caveman&lt;/code&gt; is the loudest token-saving project of 2026. Its GitHub description reads: &lt;em&gt;&amp;ldquo;Viral skill + proxy for coding agents that cuts 65% of tokens by talking like a caveman.&amp;rdquo;&lt;/em&gt; It has &lt;strong&gt;109,102 stars&lt;/strong&gt;, 6,315 forks, 71 contributors and 346,877 release-asset downloads. Its top Hacker News thread hit &lt;strong&gt;904 points and 366 comments&lt;/strong&gt; one day after the repository was created. It shipped 42 releases. JetBrains built a three-part benchmark series around it.&lt;/p&gt;
&lt;p&gt;So I cloned it at commit &lt;code&gt;8af1f1b&lt;/code&gt; (3 October 2026, 00:38Z), measured the repository, ran its own evaluation harness against its own committed data, ran its own compression benchmark against its own fixtures, read the source of its telemetry receiver and audited its CI workflows.&lt;/p&gt;</description></item></channel></rss>