The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/naver_vpn.py invokes system tools like codesign and openvpn using safe subprocess calls with argument lists. All executable paths are either hardcoded to the official application bundle or strictly validated.
[PRIVILEGE_ESCALATION]: VPN operations utilize sudo with the non-interactive flag. This ensures the AI agent cannot prompt for or capture the user's password, requiring the user to have already authorized sudo in a separate terminal session.
[SAFE]: The skill enforces a security-first posture by validating that configuration files are owned by the current user with restricted permissions (0600) and by auditing the VPN app's Team ID and Bundle ID to prevent binary impersonation.

naver-cloud-vpn