simpo-training

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

• Third-party content exposure detected (low risk: 0.10). The skill documentation references public HuggingFace datasets (such as HuggingFaceH4/ultrafeedback_binarized) for model training, but ingestion of third-party dataset content occurs via active fetching/loading rather than an unselected incoming feed.

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

• Potentially malicious external URL detected (medium risk: 0.30). The skill references GitHub repositories like huggingface/alignment-handbook via git clone, which constitutes an operational dependency retrieved from a common unofficial/third-party source.