<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Maximilian Richter — AI agent security</title><description>Maximilian Richter works on AI agent security — prompt injection, tool misuse, and how you measure whether a defense works. Software developer (Ausbildung, 2026); dual B.Sc. Computer Science at Bosch from October 2026. Endurance athlete. Based in the Allgäu.</description><link>https://richtermax.com/</link><item><title>Six lessons from AEGIS.</title><link>https://richtermax.com/blog/building-aegis/</link><guid isPermaLink="true">https://richtermax.com/blog/building-aegis/</guid><description>What building a deterministic security evaluation harness for tool-using AI agents actually taught me — about decisions, scope, and the discipline of staying boring.</description><pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Ausbildung first.</title><link>https://richtermax.com/blog/ausbildung-first/</link><guid isPermaLink="true">https://richtermax.com/blog/ausbildung-first/</guid><description>Why I did the German vocational route before going to university — and why I’m starting the university part now anyway. Practice first, theory second.</description><pubDate>Thu, 30 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Public, by accident.</title><link>https://richtermax.com/blog/public-by-accident/</link><guid isPermaLink="true">https://richtermax.com/blog/public-by-accident/</guid><description>What strangers can pull from your social media. The gap between sharing one thing and revealing another — and what to actually do about it.</description><pubDate>Mon, 20 Apr 2026 00:00:00 GMT</pubDate></item></channel></rss>