transformers

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

• Third-party content exposure detected (high risk: 0.80). This skill's examples and guidance call out AutoModel.from_pretrained and pipeline usage that automatically download models from the public Hugging Face Model Hub (https://huggingface.co/models), a repository of user-contributed/untrusted models which the agent loads and executes as part of its workflow, creating a clear vector for indirect prompt injection.