ai-dev-tools

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

• Potentially malicious external URL detected (high risk: 1.00). The skill directly runs vendor install scripts at runtime using curl|sh or curl|bash (e.g., https://claude.ai/install.sh, https://chatgpt.com/codex/install.sh, https://x.ai/cli/install.sh, https://code.kimi.com/kimi-code/install.sh, https://pi.dev/install.sh, https://gh.io/copilot-install, https://cursor.com/install, https://opencode.ai/install), which fetch and execute remote code and are presented as required install channels.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill explicitly instructs installing multiple third‑party CLIs (including running curl|bash and global npm/brew installs) which modify the host system and can access sensitive artifacts (SSH keys, cloud sessions), so it pushes the agent to change machine state and risks compromise.