querying-analytics
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill documentation provides numerous SQL templates intended for use with the
query_replo_analyticstool. The instructions clarify that the execution environment is limited to read-onlySELECToperations and that the server enforces project-level scoping using ClickHouse row policies to ensure data isolation. - [DATA_EXPOSURE]: The skill facilitates access to sensitive business metrics such as revenue and session data. It emphasizes that access is restricted to authorized project namespaces and that the agent should verify schemas and data definitions to ensure accuracy and privacy.
- [DYNAMIC_EXECUTION]: Documentation within the skill includes a JavaScript snippet designed to be manually executed by a user in a browser developer console for debugging purposes. This script is used to monitor Meta Pixel event triggers and is not executed automatically by the agent or the skill's tools.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data returned from database queries (e.g., product titles, UTM parameters). While this data is handled by the agent, the skill provides specific reasoning checks and instructions to validate definitions, reducing the risk of accidental obedience to instructions embedded in analyzed data.
Audit Metadata