<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Mcp on My Blog</title><link>https://hugo-blog.aitbytes.dev/tags/mcp/</link><description>Recent content in Mcp on My Blog</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Thu, 11 Jun 2026 12:00:00 +0000</lastBuildDate><atom:link href="https://hugo-blog.aitbytes.dev/tags/mcp/index.xml" rel="self" type="application/rss+xml"/><item><title>The MCP Attack Surface: What Your Security Team Is Missing About AI Coding Tools</title><link>https://hugo-blog.aitbytes.dev/posts/2026-06-11-mcp-attack-surface-agentic-coding/</link><pubDate>Thu, 11 Jun 2026 12:00:00 +0000</pubDate><guid>https://hugo-blog.aitbytes.dev/posts/2026-06-11-mcp-attack-surface-agentic-coding/</guid><description>&lt;p&gt;The more capable your AI coding assistant gets, the more dangerous it becomes.&lt;/p&gt;
&lt;p&gt;I know that sounds backwards. Security tools are supposed to get safer as they mature. But with agentic coding tools, the relationship between capability and risk flips in a way that nobody prepared for. Academic research published in April 2026 tested 2,000 attack instances across nine LLMs. The result? The strongest instruction-following models — the ones enterprises actually want to deploy — were the ones most likely to hand an attacker your database credentials.&lt;/p&gt;</description></item></channel></rss>