tooluniverse-dataset-discovery

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to search public repositories (Step 2: PubMed/EuropePMC, CDC/NHANES, GEO, ClinicalTrials.gov, etc.) and to download and parse arbitrary dataset URLs/APIs (Step 4: requests.get, API pagination), meaning the agent will ingest untrusted third‑party web content that can materially influence subsequent tool use and decisions.