Gemini CLI Vulnerability Allows Silent Execution of Malicious Commands on Developer Systems

Tracebit security researchers discovered a critical vulnerability in Google's Gemini CLI that allowed attackers to execute malicious commands on developers' systems through prompt injection and flawed validation.

The vulnerability exploited Gemini CLI's command execution capabilities and context file features to hide malicious code within legitimate-looking files. Attackers could embed prompt injection techniques in files like README.md, then manipulate the command whitelisting system to execute dangerous operations while obscuring the malicious activity from users through terminal output manipulation.

• Google classified this as a P1/S1 critical security issue and released a patch in version 0.1.14
• The attack combined prompt injection with improper command validation and deceptive user interface design
• Researchers successfully demonstrated credential exfiltration while keeping users completely unaware
• Multiple security researchers independently discovered similar vulnerabilities within a month of release