huggingface_transformers

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

• Third-party content exposure detected (high risk: 0.80). SKILL.md explicitly instructs the agent to load public models and datasets (e.g., AutoModel.from_pretrained("meta-llama/Llama-2-7b-hf"), AutoTokenizer.from_pretrained(...), and load_dataset("glue","mrpc")) which downloads untrusted, community-hosted content from the Hugging Face Model Hub/Datasets that the agent will consume and which can materially change inference/training behavior.