transformers

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs loading models and tokenizers from the public Hugging Face Hub (e.g., AutoModel.from_pretrained/AutoTokenizer.from_pretrained and model IDs), using datasets.load_dataset (example: "yelp_review_full"), and pipelines that accept arbitrary URLs or external files (e.g., image/audio URLs), which means it ingests untrusted, user-provided content from open third‑party sources that the agent will read and process.