molfeat

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

• Third-party content exposure detected (high risk: 0.70). The skill's documentation and workflows (ModelStore usage in SKILL.md and references) explicitly show loading pretrained models and model cards from external sources (e.g., "Access to transformer models through HuggingFace hub", PretrainedMolTransformer and store.load/model_card.usage()), meaning the agent will fetch and read untrusted third‑party model artifacts/metadata which can materially influence behavior.