tooluniverse-sdk

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

• Third-party content exposure detected (high risk: 0.80). The skill's documented workflows explicitly call public data-fetching tools (e.g., PubMed_search_articles in "Complex Workflows" and calls like UniProt_get_entry_by_accession and ChEMBL_search_molecule_by_target in SKILL.md/REFERENCE.md) and then programmatically read and use those results (e.g., extracting gene_names and driving further tool calls), so the agent ingests untrusted third‑party web/database content that can influence subsequent actions.