A Single Flaw Lets Untrusted Repos Run Code in Claude Code, Codex, Cursor, Grok

Heads up, AI coders! A critical flaw means untrusted GitHub repos can execute arbitrary code directly within tools like Claude Code and Cursor. This vulnerability could silently compromise your development environment if you open a malicious repository with these AI assistants. It's a stark reminder to vet your project sources rigorously, even when your AI is doing the heavy lifting. Stay vigilant and secure your workflow. Source: https://www.manifold.security/blog/ai-coding-agents-git-hijack