A critical command injection vulnerability in Roo-Code allows unauthenticated attackers to trigger unauthorized command execution, potentially compromising the host environment.

The application fails to properly validate inputs used by the AI agent, leading to a command injection flaw. An unauthenticated attacker can manipulate the agent to execute arbitrary system commands that bypass pre-configured allow list prefixes.

The ability to execute arbitrary system commands poses a severe risk to the integrity and confidentiality of the development environment. A successful exploit could lead to full system compromise, unauthorized data exfiltration, or the installation of persistent malicious backdoors. With a CVSS score of 8.1, this high-severity vulnerability necessitates immediate attention to prevent operational disruption and potential lateral movement within the corporate network.

Immediate Action: Update the Roo-Code extension to version 3.26.7 or later immediately to incorporate the necessary input validation fixes.

Proactive Monitoring: Review system audit logs for unexpected process execution or abnormal command-line arguments originating from the IDE environment.

Compensating Controls: Restrict the permissions of the user account running the IDE to the minimum necessary level, which limits the potential damage an attacker can inflict through command injection.

Public Exploit Available: No (exploit_available: false)

Given the potential for complete system compromise via command injection, administrators must prioritize the deployment of version 3.26.7 across all developer workstations. Organizations should verify that all instances of the Roo-Code extension are updated, as the autonomous nature of the agent increases the risk of inadvertent execution of malicious instructions. Failure to patch this vulnerability leaves development environments exposed to significant security threats.