similarity-search-patterns
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates processing external data sources and source material, which constitutes an ingestion surface for potential indirect prompt injection.
- Ingestion points: The skill reads from an "external data source" and user-provided "source material" as part of its workflow in
SKILL.md. - Boundary markers: There are no specific instructions to use boundary markers or delimiters when processing these external inputs.
- Capability inventory: The skill uses the
sandbase_runtool to execute external capabilities and model inference. - Sanitization: The instructions do not specify any sanitization or validation steps for content retrieved from external sources.
- [SAFE]: The skill is authored by the vendor of the tools it utilizes. It incorporates protective measures, such as requiring user approval and providing cost estimates before executing paid operations via the SandBase API. The workflow emphasizes local execution and minimal data sharing, reducing the overall risk profile.
Audit Metadata