OS Command Injection in SonicCloudOrg Sonic-Agent

Executive Summary

CVE-2026-15495 is a vulnerability in the Sonic Cloud Platform's sonic-agent component (version up to 2.7.2) that allows remote attackers to execute arbitrary operating system commands. This happens because the argument 'path' in the pullFile functionality is not properly sanitized and is directly used in an ADB shell command, enabling OS command injection.

The vulnerability involves two main issues: an authentication bypass due to weak secretKey exposure and token validation in the WebSocket endpoint, and a command injection flaw where malicious commands can be injected via the path parameter. Attackers can exploit this to gain remote code execution as root on the agent host.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability can have severe impacts including full system compromise. An attacker can remotely execute arbitrary commands as root on the affected device, leading to unauthorized control over the system.

• Arbitrary code execution as root on the agent host.
• Data exfiltration by writing command outputs to files and retrieving them.
• Complete device control and potential lateral movement within internal networks.

Useful 0 Not Useful 0

Detection Guidance

This vulnerability can be detected by attempting to exploit the command injection via the pullFile functionality of the Sonic Cloud Platform's sonic-agent component. The proof-of-concept (PoC) exploit script interacts with the Sonic server to log in, list agents and devices, and establish a WebSocket connection to the agent. It then injects commands into the pullFile parameter to execute arbitrary OS commands.

To detect the vulnerability, you can use the PoC Python script from Resource 1 which provides an interactive shell to execute commands remotely on the target system. This script targets the /server/api/controller endpoint and requires the target to be running a vulnerable version of sonic-agent.

Suggested commands include using the PoC to inject simple OS commands such as 'id' or 'whoami' via the pullFile parameter to verify if command execution is possible. The script automates this process, but manual testing could involve sending crafted WebSocket messages with injected commands in the path argument.

Useful 0 Not Useful 0

Mitigation Strategies

Immediate mitigation steps include restricting access to the sonic-agent service to trusted users and networks, since the vulnerability allows remote command execution due to authentication bypass and command injection.

Additional steps involve disabling or isolating the vulnerable sonic-agent component if possible, especially since the affected products are no longer supported by the maintainer.

Longer-term mitigations recommended by the vendor or community include applying strict path validation, removing shell interpretation from command execution, enhancing secretKey and token validation, and running the agent as a non-root user to limit impact.

Useful 0 Not Useful 0

Compliance Impact

The vulnerability in SonicCloudOrg sonic-agent allows remote attackers to execute arbitrary OS commands as root, potentially leading to full system compromise, data exfiltration, and unauthorized device control.

Such unauthorized access and data breaches could result in non-compliance with common standards and regulations like GDPR and HIPAA, which mandate protection of personal and sensitive data against unauthorized access and disclosure.

However, the provided information does not explicitly discuss the impact on compliance with these standards or any regulatory implications.

Useful 0 Not Useful 0