lora

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

• Third-party content exposure detected (high risk: 0.90). The skill directly loads public models, adapters, and datasets from Hugging Face (e.g., AutoModelForCausalLM.from_pretrained("meta-llama/..."), PeftModel.from_pretrained, model.push_to_hub, and load_dataset("tatsu-lab/alpaca")), which are open, user-provided web resources the agent would ingest and thus could carry indirect prompt injections.