hetzner-server

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs using user-data that curls and pipes a public GitHub raw URL (https://raw.githubusercontent.com/connorads/dotfiles/master/install.sh | bash) and to clone arbitrary GitHub repos, meaning the skill fetches and executes untrusted, user-generated third-party content (run as root via cloud-init) that can materially change behavior.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill instructs making persistent system changes (updating ~/.ssh/config and ~/.ssh/known_hosts), running sudo commands via SSH, and using root user-data that creates users and modifies system files on managed servers, which alters machine state and matches flagged behaviors (e.g., modifying SSH config and creating users).