devops

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

• Third-party content exposure detected (high risk: 1.00). The skill's required workflows explicitly fetch and act on open third-party content (e.g., the one‑line installer curl|bash in references/BUNDLED_SCRIPTS.md for coolify, cloudflared downloads from GitHub releases in references/cloudflare-tunnel.md, and provider/API interactions in references/contabo.md that parse API responses to drive provisioning), so the agent will ingest and act on untrusted public content that can materially influence subsequent tool use and decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes runtime fetch-and-execute commands that download and run remote scripts (e.g. curl -fsSL https://raw.githubusercontent.com/YOUR-ORG/vibeskills-demo/main/.claude/skills/devops/scripts/coolify-enhanced-setup.sh | bash and curl -fsSL https://cdn.coollabs.io/coolify/install.sh | bash and GitHub release downloads for cloudflared), which clearly fetch external code used/required at runtime and execute it, creating a high-risk remote code execution vector.