alterlab-zinc-db

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

• Third-party content exposure detected (high risk: 1.00). This skill programmatically fetches and ingests data from public ZINC endpoints (e.g., https://cartblanche22.docking.org/ and https://files.docking.org/zinc22/ via the many curl/wget examples), and the returned untrusted public data (SMILES, tranche, catalogs) is parsed and used to drive further actions (filtering, downloading tranche files, preparing docking libraries), so third‑party content can materially influence tool use and next steps.