dl-transformer-finetune

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs runtime loading of public third-party artifacts (e.g., datasets.load_dataset("multi_nli") and AutoModel/AutoTokenizer.from_pretrained("bert-base-uncased", "meta-llama/...")) which ingest models and datasets from the Hugging Face/public hubs (user-contributed content) that will be read and used in the fine-tuning workflow and can materially change model behavior.