linux-cron-service-abuse

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly requires the agent to "leverage existing credentials" and to include "New credentials or tokens found" in its return summary (and save evidence), which forces the LLM to handle and potentially output secret values verbatim, creating an exfiltration risk.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). Suspicious — the localhost URLs are Docker/host API endpoints (commonly abused via a writable docker socket to spawn privileged containers or execute commands) and the two GitHub links are public PwnKit exploit repositories (PoC code/binaries for local privilege escalation), so together they are high-risk vectors for distributing or running malware.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This skill file is explicitly a high-risk offensive playbook: it provides step-by-step instructions to obtain root privileges, create persistent backdoors (SUID root shells, reverse shells, xinetd/backdoor services, systemd/crontab modifications), perform remote code execution via D-Bus and Unix sockets, exploit known privilege-escalation CVEs, and abuse docker sockets — all patterns that enable unauthorized access, persistence, credential theft, and data exfiltration.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The prompt explicitly instructs the agent to perform privilege-escalation actions that modify system files (cron, systemd units, init scripts, service binaries, D-Bus policies), create SUID root binaries and new user accounts, and bypass security mechanisms (polkit exploits, Docker socket abuse), which directly compromise the host state.