diagnose

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

• Third-party content exposure detected (high risk: 0.70). The skill's diagnostics explicitly check and download models from the public HuggingFace cache (see SKILL.md and references/common-issues.md: "ls ~/.cache/huggingface/hub/models--..." and "curl https://huggingface.co/"), which are third‑party, community-hosted assets the workflow ingests and depends on, so untrusted external content can influence runtime behavior.