omicverse-micro-metabol-paired
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes the
ov.micro.fetch_franzosa_ibd_2019function, which downloads scientific datasets from external repositories to a local directory. This is a standard and expected feature for scientific analysis skills providing demonstration data.\n- [PROMPT_INJECTION]: The skill is identified as having an attack surface for indirect prompt injection due to its ingestion of external data.\n - Ingestion points: Biological datasets are ingested via the
ov.micro.fetch_franzosa_ibd_2019function as specified inSKILL.md.\n - Boundary markers: The instructions do not define clear delimiters or use specific language to prevent the agent from executing instructions potentially embedded in the ingested data.\n
- Capability inventory: The skill uses
torchfor model training and performs local file system writes to a cache directory, which represent the available capability set for potential exploitation.\n - Sanitization: There is no documentation of sanitization or validation processes applied to the external CSV files before they are processed by the integrated models.
Audit Metadata