hugging-face-model-trainer

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow can ingest outsider-authored free text from public Hugging Face datasets/models via datasets.load_dataset(...) / from_pretrained(...) (e.g., trl-lib/Capybara, trl-lib/math_shepherd, or user-supplied dataset IDs), and those dataset examples’ text fields are fed into the training pipeline/LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs remote scripts at runtime (e.g., "https://huggingface.co/datasets/mcp-tools/skills/raw/main/dataset_inspector.py" invoked via hf_jobs and the convert script clones and builds "https://github.com/ggerganov/llama.cpp.git" in scripts/convert_to_gguf.py), which fetches and executes external code that the workflow depends on.