<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Token-Optimization on Tobias Theel | Senior Software Engineer</title><link>https://blog.noobygames.de/tags/token-optimization/</link><description>Recent content in Token-Optimization on Tobias Theel | Senior Software Engineer</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 13 Nov 2026 08:00:00 +0100</lastBuildDate><atom:link href="https://blog.noobygames.de/tags/token-optimization/index.xml" rel="self" type="application/rss+xml"/><item><title>Token Cost Minimization in Agentic AI — Part 1: Where the Tokens Go, Caveman and Ponytail</title><link>https://blog.noobygames.de/blog/token-cost-1-caveman-ponytail/</link><pubDate>Fri, 13 Nov 2026 08:00:00 +0100</pubDate><guid>https://blog.noobygames.de/blog/token-cost-1-caveman-ponytail/</guid><description>&lt;blockquote&gt;&#10;&lt;p&gt;&lt;strong&gt;Series: Token Cost Minimization in Agentic AI.&lt;/strong&gt; Part 1: where the tokens go, Caveman and Ponytail (you are here) · &lt;span class="link-upcoming"&gt;Part 2: code-review-graph and hooks&lt;/span&gt; · &lt;span class="link-upcoming"&gt;Part 3: Headroom, auto-updates and the combined setup&lt;/span&gt;&lt;/p&gt;&#10;&lt;/blockquote&gt;&#10;&lt;p&gt;Coding agents like &lt;a href="https://code.claude.com/docs/en/overview" rel="noopener"&gt;Claude Code&lt;/a&gt;, Codex or Cursor are billed per token, or they burn through a plan&amp;rsquo;s usage limit per token. A long agent session easily reads millions of tokens. A whole ecosystem of tools now promises to cut that bill by &amp;ldquo;50 to 90 percent&amp;rdquo;.&lt;/p&gt;</description></item></channel></rss>