omni-to-snowflake-semantic-view
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes legitimate CLI tools to manage configurations and query data from external platforms.\n
- Evidence: Uses the Omni CLI (
omni) and Snowflake tools (snow,snowsql) for model exploration and SQL execution.\n- [EXTERNAL_DOWNLOADS]: The skill suggests installing a standard integration library for Snowflake connectivity.\n - Evidence: Recommends the installation of
snowflake-connector-pythonvia pip.\n- [DYNAMIC_EXECUTION]: The skill generates SQL statements by mapping and concatenating metadata from external YAML definitions into database-ready expressions.\n - Execution method: SQL execution through terminal-based CLIs or Python scripts.\n
- Patterns: Translates Omni dimension and measure logic into SQL
CASE WHENandCOALESCEexpressions for Snowflake Semantic Views.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external metadata files that serve as the foundation for code generation, presenting a potential injection surface.\n - Ingestion points: Data extracted from Omni topic, view, and relationship YAML files via
omni models yaml-get(SKILL.md).\n - Boundary markers: None specified to delimit or sanitize the imported field definitions.\n
- Capability inventory: Shell command execution and SQL query processing.\n
- Sanitization: No explicit validation of field names or SQL expressions provided in the source YAML is detailed in the conversion logic.
Audit Metadata