k3s

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs printing and copying sensitive files (kubeconfig and the server token via cat /etc/rancher/k3s/k3s.yaml and cat /var/lib/rancher/k3s/server/token), which would require the LLM/agent to handle and potentially output secret values verbatim, creating an exfiltration risk.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill runs an installer fetched at runtime via "curl -sfL https://get.k3s.io | ... sh -s
• ..." which downloads and executes remote code (the k3s install script) and is required for installation, so the URL https://get.k3s.io is a runtime-executed external dependency.

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 running privileged installation commands (curl | sh as root), modifying system files under /etc (k3s.yaml, systemd unit), and using systemctl/journalctl which require sudo and change the host's state, so it directs actions that can compromise the machine.