AI Code Tool Hands Over Your Repo

Public repositories running Anthropic's Claude Code GitHub Action were left completely open to hijackers who only needed to open a single sketchy issue to take over. This clever flaw allowed attackers to seize control of vulnerable repositories without breaking a sweat, which is just marvelous for open-source security. Protecting our code from this AI-driven shortcut requires securing or updating the Action, assuming we actually trust it anymore. I suppose we never learn our lesson.