The vulnerability resides in the detect_dangerous_command function of terminal_tool within the NousResearch hermes-agent. The function incorrectly handles user-supplied input, allowing an attacker to inject arbitrary OS commands, which may compromise confidentiality, integrity, or availability. The flaw carries a CVSS score of 6.9, indicating a moderate severity, and the exploit has been publicly disclosed and may be used by attackers.

Affected systems include all installations of NousResearch hermes-agent up to and including the commit identifier 5157f5427f19488b31c6fdebbacd15d798ce7f63. The entry point for exploitation is the detect_dangerous_command function within tools/approval.py of the terminal_tool component.

The CVSS score of 6.9 reflects moderate risk, and the EPSS score is 1%. The vulnerability is not listed in the CISA KEV catalog, yet its remote nature and public disclosure make exploitation feasible from outside the system. Attackers may craft input to trigger arbitrary command execution, therefore remediation is strongly advised to prevent potential compromise.