<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>OWASP on Bruce on AI Engineering</title><link>http://www.heyuan110.com/tags/owasp/</link><description>Recent content in OWASP on Bruce on AI Engineering</description><generator>Hugo</generator><language>en</language><lastBuildDate>Mon, 23 Feb 2026 10:00:00 +0800</lastBuildDate><atom:link href="http://www.heyuan110.com/tags/owasp/index.xml" rel="self" type="application/rss+xml"/><item><title>MCP Security Guide: Attack Patterns, Real CVEs, and Defense Strategies for AI Agents</title><link>http://www.heyuan110.com/posts/ai/2026-02-23-mcp-security-guide/</link><pubDate>Mon, 23 Feb 2026 10:00:00 +0800</pubDate><guid>http://www.heyuan110.com/posts/ai/2026-02-23-mcp-security-guide/</guid><description>&lt;p&gt;&lt;strong&gt;518 official MCP Servers, 41% lacking authentication.&lt;/strong&gt; This is not a hypothetical threat model — it is real data from a February 2026 security audit.&lt;/p&gt;
&lt;p&gt;The MCP (Model Context Protocol) registry exploded from 90 servers to 518 in just one month. The ecosystem is expanding far faster than its security infrastructure can keep up. While developers eagerly plug MCP Servers into their AI agents, attackers are watching the same door.&lt;/p&gt;</description></item></channel></rss>