AI Skills Hide Malicious Code in Test Files—Here’s How It Works
A new security flaw reveals how malicious code embedded in test files bypasses AI agent scanners entirely. Here’s why developers must rethink their threat models before trusting AI Skills.