Critical flaw in Claude Code, OpenAI Codex, GitHub Copilot, and Gemini CLI
heise online

Picture: heise online
The four coding agents Claude Code, OpenAI Codex, GitHub Copilot, and Gemini CLI are affected by the Zero-Click RCE (Remote Code Execution) vulnerability Plugin4Shell, which injects malicious code into developers' computers. This happens via manipulated marketplace plug-ins that the agents automatically load. According to the AI security firm AIR, primarily the providers of the coding agents can fix the security vulnerability, as the marketplaces have no influence on what the coding agent does with the downloaded code. Anthropic and OpenAI have already delivered and patched their agents. However, there is no fix for Copilot yet, and the consumer version of Gemini CLI remains permanently vulnerable. A case where the security vulnerability was actually exploited is not known, however.
This report was first published by heise online. What you see here is a brief excerpt. The full text is theirs.
Continue reading at heise online


