Three AI coding agents leaked secrets through a single prompt injection. On
<p>A security researcher, working with colleagues at <a href="https://www.jhu.edu/">Johns Hopkins University</a>, opened a GitHub pull request, typed a malicious instructio [...]