<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>AI Agent Security blog</title><description>Research and field notes on securing enterprise AI agents.</description><link>https://ai-agentsecurity.com/</link><language>en</language><item><title>How AI Agents Really Work Under the Hood, Frame by Frame</title><link>https://ai-agentsecurity.com/blog/ai-agent-security-lab/</link><guid isPermaLink="true">https://ai-agentsecurity.com/blog/ai-agent-security-lab/</guid><description>How an AI agent works under the hood: a Python loop, an MCP server, an LLM and 25 Splunk events read frame by frame. Code snippets and a lab to rebuild it.</description><pubDate>Thu, 08 Oct 2026 00:00:00 GMT</pubDate><category>AI agents</category><category>MCP security</category><category>Agent logging</category><category>Splunk</category></item></channel></rss>