nvidia-ontology-management
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill manages untrusted data ingestion via YAML model imports and SQL expression validation endpoints.
- Ingestion points: Untrusted content enters the agent context through
/api/model/importand/api/sql-attributes/validate(as documented in SKILL.md and references/write-api.md). - Boundary markers: The instructions do not specify explicit delimiters for the agent when processing these external strings.
- Capability inventory: The skill allows structural modification of the semantic catalog, including patching terms and creating SQL derivations.
- Sanitization: The risk is mitigated by mandatory SQL validation, explicit user consent for destructive imports (replace=true), and a rigorous readback workflow comparing pre- and post-state hashes to ensure parity and detect tampering.
Audit Metadata