tooluniverse

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and processes content from public third-party sources (e.g., PubMed via "PubMed_search", GEO via "GEO_download_dataset", and databases like OpenTargets/UniProt/PDB shown in the workflows and references) and then feeds that retrieved text/data into LLM summarization and downstream analysis, so untrusted external content could carry indirect prompt injections.