<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai on Thomas Peters</title><link>https://www.sirmysterion.com/tags/ai/</link><description>Recent content in Ai on Thomas Peters</description><generator>Hugo</generator><language>en</language><managingEditor>thomas@tnepeters.com (Thomas Peters)</managingEditor><webMaster>thomas@tnepeters.com (Thomas Peters)</webMaster><copyright>&lt;a href="https://creativecommons.org/licenses/by-nc/4.0/" target="_blank" rel="noopener"&gt;CC BY-NC 4.0&lt;/a&gt;</copyright><lastBuildDate>Tue, 14 Jul 2026 10:02:14 -0600</lastBuildDate><atom:link href="https://www.sirmysterion.com/tags/ai/index.xml" rel="self" type="application/rss+xml"/><item><title>An AI Agent Lives in My Cluster: Read-Only RBAC, Network Policies, and Getting Red-Teamed by Friends</title><link>https://www.sirmysterion.com/posts/20260713-hermes-agent-in-cluster/</link><pubDate>Mon, 13 Jul 2026 00:00:00 +0000</pubDate><author>thomas@tnepeters.com (Thomas Peters)</author><guid>https://www.sirmysterion.com/posts/20260713-hermes-agent-in-cluster/</guid><description>&lt;h1 id="the-problem"&gt;The Problem&lt;/h1&gt;
&lt;p&gt;I run &lt;a href="https://github.com/SirMysterion/hermes-agent"&gt;Hermes&lt;/a&gt; - a persistent, tool-using AI agent - inside my Kubernetes cluster.
It helps me operate the homelab: check on failing pods, read ArgoCD app status, dig through CloudNativePG backups, tell me why a node is unhappy.
To do any of that it needs &lt;code&gt;kubectl&lt;/code&gt;, cluster credentials, and network reach.&lt;/p&gt;
&lt;p&gt;Which means I deliberately deployed a curious, tool-wielding, LLM-driven process &lt;em&gt;inside&lt;/em&gt; my cluster, gave it a service account, and then - because apparently I like living dangerously - gave my friends their own instances.&lt;/p&gt;</description></item></channel></rss>