alterlab-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 (SKILL.md and references/api_reference.md and references/available_featurizers.md) explicitly describes using ModelStore.load() and PretrainedMolTransformer to fetch and load pretrained models from public hubs (e.g., HuggingFace and other external model collections), meaning the agent will ingest untrusted, third‑party model artifacts and model cards as part of its runtime workflow which can materially affect downstream decisions and behavior.