code-review

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). This is an unverified third‑party GitHub repository being used as a direct install source for a CLI extension (gemini extensions install), which can execute arbitrary code when installed and therefore is a potential malware distribution vector.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's CI and hook examples execute third‑party packages/actions at runtime (e.g., the GitHub Action "openai/codex-action@main" and runtime npx/npm installs like "npx @anthropic-ai/claude-code" and "npm install -g @openai/codex" which fetch and run remote code), so these external references are runtime dependencies that execute remote code.