transformers

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

• Third-party content exposure detected (high risk: 0.80). This skill explicitly loads public models and datasets (e.g., AutoModel.from_pretrained, pipeline(..., model=...), and datasets.load_dataset("imdb")) and references the Hugging Face Model Hub and Datasets Hub, which are public/user-contributed sources the agent ingests and would read/interpret.