<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>thephreck</title><link>https://phreck.com/</link><description>Recent content on thephreck</description><generator>Hugo</generator><language>en</language><lastBuildDate>Tue, 21 Jul 2026 07:26:59 -0700</lastBuildDate><atom:link href="https://phreck.com/index.xml" rel="self" type="application/rss+xml"/><item><title>From Autocomplete to Autonomy: Leveling Up in the Agentic Era</title><link>https://phreck.com/posts/ai-context-and-workflows/</link><pubDate>Tue, 21 Jul 2026 07:26:59 -0700</pubDate><guid>https://phreck.com/posts/ai-context-and-workflows/</guid><description>A practical guide to moving from basic AI autocomplete to autonomous agentic workflows using context engineering, custom agents, and multi-agent handoffs.</description></item><item><title>Building contain-yourself: A Dockerized MCP Sandbox for AI Security Agents</title><link>https://phreck.com/projects/contain-yourself/</link><pubDate>Wed, 15 Jul 2026 12:15:11 +0000</pubDate><guid>https://phreck.com/projects/contain-yourself/</guid><description>How I wrapped standard Linux networking utilities in Model Context Protocol (MCP) servers and Docker containers to give LLMs safe, structured access to recon tools.</description></item><item><title>The AI Rabbit Hole</title><link>https://phreck.com/posts/the-ai-rabbit-hole/</link><pubDate>Tue, 24 Feb 2026 08:52:17 +0000</pubDate><guid>https://phreck.com/posts/the-ai-rabbit-hole/</guid><description>Over the past 18 months or so, I&amp;rsquo;ve gone deep down the AI rabbit hole. Not the “use chatgpt” rabbit hole, not the “call openai APIs for my webapp” rabbit hole, but the deep, wide and evergreen open source model, with local compute rabbit hole.</description></item><item><title>Skills are Code: Why Agentic Workflows Need to Be Treated Like Software</title><link>https://phreck.com/posts/skills-are-code/</link><pubDate>Wed, 18 Feb 2026 08:52:17 +0000</pubDate><guid>https://phreck.com/posts/skills-are-code/</guid><description>Skills should be treated as actual units of code: reusable, maintainable, and version-controlled, not static, copy-pasted text files.</description></item><item><title>A Homelab Raspberry Pi Cluster</title><link>https://phreck.com/projects/raspberry-pi-cluster-build/</link><pubDate>Thu, 10 Oct 2024 21:29:39 -0700</pubDate><guid>https://phreck.com/projects/raspberry-pi-cluster-build/</guid><description>I decided to take the plunge and build a nifty Raspberry Pi cluster. Lets take a journey.</description></item><item><title>About Me</title><link>https://phreck.com/about/</link><pubDate>Fri, 20 Oct 2023 10:00:00 -0700</pubDate><guid>https://phreck.com/about/</guid><description>&lt;div class="flex flex-col sm:flex-row items-center gap-6 mb-8"&gt;
 
 
 
 &lt;div class="w-32 h-32 rounded-full border-2 border-primary/50 overflow-hidden shadow-lg flex-shrink-0 my-0"&gt;
 &lt;img src="https://phreck.com/images/phreck_logo.png" alt="phreck" class="w-full h-full object-cover my-0" style="border-radius: 9999px !important;" /&gt;
 &lt;/div&gt;
 
 
 &lt;div&gt;
 
 &lt;p class="text-xl font-semibold text-foreground my-0"&gt;Hey, I&amp;#39;m phreck! 👋&lt;/p&gt;
 
 
 &lt;p class="text-base text-foreground/80 mt-2 mb-0"&gt;I am a hacker and tech nerd based in the San Francisco Bay Area. I enjoy tinkering with code, homelabs, open source, decentralized technologies, and making physical things&lt;/p&gt;
 
 &lt;/div&gt;
&lt;/div&gt;

&lt;hr&gt;
&lt;div class="prose prose-sm dark:prose-invert max-w-none text-foreground/80"&gt;
 &lt;h4 id="-homelab--over-engineering"&gt;🖥️🛠️ Homelab &amp;amp; Over-Engineering&lt;/h4&gt;
&lt;p&gt;I don&amp;rsquo;t just host standard home servers; I host a disproportionate amount of enterprise plumbing in my closet. I run a full Wi-Fi 7 network stack throughout the house, backed by a 10GbE fiber backbone connecting my compute clusters and storage pools, with 2.5GbE run out to the edge. If it can be dockerized, containerized, or run in a Kubernetes pod, it lives here. Currently running local Ollama and vLLM instances because hosting my own deep reasoning setups is far more fun (and loud) than renting cloud instances.&lt;/p&gt;</description></item></channel></rss>