AI coding agents' 0-click RCE flaw could hand attackers keys to the kingdom
The major AI coding agents, including Anthropic's Claude Code, OpenAI's Codex, Google's Gemini CLI, Microsoft's Copilot, and GitHub Copilot, have a zero-click vulnerability that allows remote code execution, giving attackers potentially unfettered access to sensitive data and assets.
The exploit, dubbed 'Plugin4Shell,' affects all of the major coding agents and could give attackers full access to every asset and piece of data that the agent can reach, researchers warn, highlighting the severity of the issue.
The vulnerability, a 'plugin SHA-pinning bypass,' sits in how agents enforce marketplaces' SHA-pinning mechanism, which locks agent plugins and skills to a specific, immutable commit hash, allowing attackers to exploit the flaw and execute malicious code without user interaction.
Story 1 of 10