A developer embedded a prompt injection attack into jqwik, a Java testing framework. The hidden instruction targeted AI coding agents, instructing them to delete app output, highlighting a novel security vulnerability for AI-assisted development workflows.
Opening Kapyn…