The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/agent.py uses subprocess.run to execute system commands such as wevtutil and powershell. The implementation is vulnerable to command injection because it does not sanitize user-supplied hostnames before passing them to the shell.\n- [DYNAMIC_EXECUTION]: In scripts/agent.py, the check_smb_signing function dynamically constructs PowerShell command strings by interpolating variables from command-line arguments. An attacker providing a malicious hostname (e.g., containing single quotes and semicolons) can execute arbitrary PowerShell commands.\n- [PRIVILEGE_ESCALATION]: The skill requires Administrative or Domain Admin privileges to function correctly, as it must query the Windows Security Event Log and audit remote SMB configurations. Executing this skill in a high-privilege context increases the impact of the identified command injection vulnerability.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Windows Event Logs which could be influenced by an attacker to manipulate analysis results. \n
Ingestion points: Event Log queries in scripts/agent.py. \n
Boundary markers: None. \n
Capability inventory: subprocess.run for log querying and PowerShell execution. \n
Sanitization: None.

hunting-for-ntlm-relay-attacks