<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>sentinel on naved // security notes</title><link>https://naved124.github.io/tags/sentinel/</link><description>Recent content in sentinel on naved // security notes</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sun, 08 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://naved124.github.io/tags/sentinel/index.xml" rel="self" type="application/rss+xml"/><item><title>sentinel-detections: KQL detection rules that have to prove themselves</title><link>https://naved124.github.io/notes/building-sentinel-detections/</link><pubDate>Sun, 08 Nov 2026 00:00:00 +0000</pubDate><guid>https://naved124.github.io/notes/building-sentinel-detections/</guid><description>&lt;h2 id="the-idea-in-one-paragraph"&gt;The idea in one paragraph&lt;/h2&gt;&#10;&lt;p&gt;A detection rule is a claim: &amp;ldquo;if this attack happens, I will fire, and if it doesn&amp;rsquo;t, I&amp;rsquo;ll stay&#10;quiet.&amp;rdquo; Most rules never get checked against either half of that claim until a real incident or&#10;a flood of false positives does it for you. &lt;a href="https://github.com/Naved124/sentinel-detections"&gt;sentinel-detections&lt;/a&gt;&#10;is fifteen Microsoft Sentinel analytics rules written in KQL, where every rule ships with test&#10;cases for both halves: attack data it must alert on, and the closest normal behaviour it must&#10;ignore. Every rule also has a triage playbook, because an alert nobody knows how to investigate&#10;is just noise with a severity label.&lt;/p&gt;</description></item></channel></rss>