managing-huggingface

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

• Third-party content exposure detected (high risk: 0.75). At runtime the skill calls Hugging Face Hub public endpoints (e.g., hf_api "models/${MODEL_ID}", datasets/${DATASET_ID}, spaces/${SPACE_ID}, and datasets/${DATASET_ID}/tree/main) and then passes the returned JSON fields (including description/metadata) through jq into the agent’s output/LLM context, which originates from outsider-authored model/dataset/space content on the public web.