tcga-bulk-data-preprocessing-with-omicverse
Warn
Audited by Snyk on Jul 22, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.75). High likelihood: the required workflow loads TCGA clinical files (from a user-provided
clinical.cart.<date>/directory) and sample sheet/archives from GDC downloads; these outsider-authored texts (clinical XML/JSON and TSV) are parsed into runtime Python objects and thus become LLM-readable context via the agent’s “confirm paths / run code” narrative and any surfaced parsing/validation errors/logs.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata