gemma4-local-deploy

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

• Potentially malicious external URL detected (high risk: 0.80). Commands in the skill explicitly fetch model artifacts at runtime (e.g., llama-server -hf ggml-org/gemma-4-12B-it-GGUF:Q4_K_M and huggingface-cli download google/gemma-4-12B-it-qat-q4_0-gguf), which will change the agent's behavior by loading remote model weights required for operation.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). Flagged because the skill explicitly directs installing or upgrading software, stopping other processes, overwriting or deleting models/configuration, and changing firewall/listening behavior — all actions that modify system state and may require privileged escalation.