tooluniverse-single-cell

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md workflow explicitly calls ToolUniverse tools (e.g., tu.run_tool("OmniPath_get_ligand_receptor_interactions", databases="CellPhoneDB,CellChatDB") and other external annotation tools like MyGene/Ensembl/UniProt), which fetches public third-party database content that the agent ingests and uses to score interactions and drive downstream analysis, so untrusted external data can materially influence actions.