hebrew-llm-eval-suite

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow uses load_dataset to fetch HuggingFace benchmark datasets (HeQ, HebrewSentiment, HebNLI) whose text content originates from public corpora such as Hebrew Wikipedia, Geektime, and consumer surveys, representing outsider-authored content that is automatically ingested at runtime without prior item selection.