matchms

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly loads spectra from public repositories via load_from_usi (references/importing_exporting.md) and fetches structure annotations from PubChem via derive_annotation_from_compound_name (references/filtering.md), meaning it ingests untrusted, user-generated third‑party content (e.g., GNPS/USI entries) that is parsed and used to drive processing and matching decisions.