cubes
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data retrieved from the RevOS MCP server (resources: cubes, tables, segments, and connections).
- Ingestion points: Data is ingested through
api_readcalls described inSKILL.mdand the reference documentation. - Boundary markers: The instructions lack explicit delimiters or instructions for the agent to ignore potentially malicious content embedded within the retrieved data.
- Capability inventory: The agent can create, edit, and delete model definitions (
api_write,api_delete) and execute SQL queries (api_readwithquerymethod). - Sanitization: No explicit sanitization or validation steps are defined for the metadata and SQL snippets retrieved from the API before they are used in subsequent operations.
Audit Metadata