rowlab-model-selection
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest and analyze rows from external datasets, which presents a surface for indirect prompt injection if the data contains malicious content designed to alter agent behavior.
- Ingestion points: Dataset rows and sampled data files used for model comparison.
- Boundary markers: The instructions lack specific markers or delimiters to isolate untrusted data from the agent's instructions.
- Capability inventory: The skill describes data manipulation (SQL-style sampling), model comparison tools (columns_compare), and the retrieval of external API documentation.
- Sanitization: There are no instructions for sanitizing or validating the contents of the rows before they are processed by the agent.
Audit Metadata