This vulnerability in NousResearch hermes-agent (versions =2026.4.0 through =2026.4.16) involves the execute_code function of the tools/code_execution_tool.py file in the Environment Variable Handler component. The issue results in a sandbox bypass or sandbox issue, allowing remote attackers to potentially execute code outside the intended restricted environment. Exploits are publicly available, increasing the risk of exploitation. The vendor has not provided any response or remediation guidance, and no patch or official fix is currently documented.

The vulnerability allows remote attackers to bypass sandbox restrictions in the hermes-agent, potentially enabling unauthorized code execution or manipulation outside the intended sandbox environment. This could lead to compromise of the host system or unauthorized actions performed by the agent. The CVSS 4.0 base score is 6.9 (medium severity), reflecting network attack vector, low complexity, no privileges or user interaction required, and limited confidentiality, integrity, and availability impacts.

No official fix or patch is currently available from the vendor, and the vendor has not responded to disclosure. Users should monitor vendor communications for updates. Until a patch is released, consider restricting network exposure of the hermes-agent service and applying compensating controls to limit potential exploitation. Patch status is not yet confirmed — check the vendor advisory for current remediation guidance.