gguf-quantization

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly instructs downloading and converting models from public sites (e.g., "huggingface-cli download meta-llama/..." and links to https://huggingface.co/TheBloke and cloning https://github.com/ggml-org/llama.cpp), which are untrusted, user-hosted third-party sources whose model files would be loaded and interpreted as part of the conversion/quantization workflow.