cli-anything

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

• Third-party content exposure detected (high risk: 0.80). The skill's required execution flow explicitly fetches and runs public third-party code (e.g., "curl -fsSL https://ollama.com/install.sh | sh" for Ollama and "git clone" for ComfyUI/Stable Diffusion) as part of normal operation, which ingests untrusted public content that can materially affect tooling and behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs runtime install commands that fetch-and-execute remote code — e.g. "curl -fsSL https://ollama.com/install.sh | sh" (and runtime git-clone + pip install for ComfyUI / AUTOMATIC1111 stable-diffusion), so external content would be executed as a required dependency.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly instructs the agent to auto-install software system-wide (including commands like "sudo apt-get install", winget/brew installs, curl|sh installers, and service-related packages), and to run privileged install steps and create harnesses — actions that modify the machine's state and require elevated privileges.