huggingface-jobs

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime path submits user-provided script strings/URLs to the hf_jobs() job runner (e.g., jobs/SKILL.md “submit jobs using hf_jobs() … The script parameter accepts Python code directly”), so an outsider can inject arbitrary free text that the job executes/ingests without first selecting a specific trusted item.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes explicit HF Jobs examples that pass remote script URLs to hf_jobs()/uv run which will fetch and execute remote code at runtime (e.g., https://huggingface.co/datasets/uv-scripts/synthetic-data/raw/main/cot-self-instruct.py), so these URLs can directly control executed code.