<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Spotlighting on Nitin Kumar Singh</title><link>https://nitinksingh.com/tags/spotlighting/</link><description>Recent content in Spotlighting on Nitin Kumar Singh</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Nitin Kumar Singh. All rights reserved.</copyright><lastBuildDate>Tue, 04 Aug 2026 08:00:00 +0530</lastBuildDate><atom:link href="https://nitinksingh.com/tags/spotlighting/index.xml" rel="self" type="application/rss+xml"/><item><title>Defense-in-Depth for AI Agents on Azure: Prompt Shields, Spotlighting, and the Coverage Gap</title><link>https://nitinksingh.com/posts/prompt-shields-spotlighting-defense-in-depth/</link><pubDate>Tue, 04 Aug 2026 08:00:00 +0530</pubDate><guid>https://nitinksingh.com/posts/prompt-shields-spotlighting-defense-in-depth/</guid><description>Prompt Shields is a real input classifier, but it&amp;rsquo;s probabilistic — and Spotlighting and Groundedness are model-only, so an agent&amp;rsquo;s tool calls can pass completely unscanned. Here&amp;rsquo;s what each defense blocks, where the agent coverage gap is, and how to layer probabilistic filters with deterministic controls.</description></item></channel></rss>