vm-security
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell scripts (
audit.shandharden.sh) on remote servers via SSH. These scripts are designed for system administration and security auditing. - Evidence:
ssh $CM USER@HOST 'sudo bash -s -- --known-ips IPS' < scripts/audit.shinSKILL.md. - [DATA_EXFILTRATION]: The skill optionally sends SSH login notifications to a user-provided Slack or Discord webhook.
- Evidence: The
--webhookparameter inharden.shconfigures apam_execscript to send login events to the specified URL. - The implementation uses a backgrounded
curlcommand with a short timeout to ensure system performance is not impacted by network latency. - [SAFE]: The skill follows security best practices by encouraging the use of multiplexed SSH connections, verification of new admin users before locking out root, and providing a clear rollback mechanism. The
audit.shscript is strictly read-only and does not modify the target system.
Audit Metadata