agentiko-hermes

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill doc embeds an explicit dashboard username/password ("admin / agentiko"), which requires the LLM to reproduce a secret credential verbatim if it describes or instructs login steps, creating an exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The document clearly describes tooling and workflows for automated offensive operations — remote command execution via an SSH worker, automated/recurring scans with results delivered to Telegram, WAF-evasion techniques, explicit exploitation guidance, and an approval-bypass "YOLO" mode — indicating deliberate malicious intent and capabilities for data exfiltration and remote abuse.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs how to disable command-approval safeguards (YOLO), how to persistently turn approvals off, and describes cron jobs that run with YOLO-equivalent behavior and script execution (including writing/executing .sh files), which directly bypasses security controls and enables automated state-changing actions on the host.