deepchem

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

• Third-party content exposure detected (high risk: 0.70). This skill explicitly loads and processes open/public third‑party datasets (e.g., MoleculeNet via dc.molnet.load_tox21/load_bbbp) and arbitrary user-provided files through CSVLoader/SDFLoader/FASTALoader, which the agent is expected to read and interpret as part of its workflows, exposing it to untrusted external content.