carto-create-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill facilitates legitimate data orchestration tasks within the CARTO ecosystem, relying on official MCP and CLI tools. There are no indications of unauthorized data access, obfuscation, or persistence techniques.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection as it processes data from external warehouse sources.\n
- Ingestion points: Untrusted data enters the agent context via exploration tools like
explore_dataandcarto connections describeduring the initial discovery phase.\n - Boundary markers: Mitigations include a mandatory phase for user confirmation (Phase 3) and the use of structural and remote validation tools (
validateandverify-remote) to check workflow integrity.\n - Capability inventory: Capabilities include creating, updating, running, and scheduling workflows, as well as executing SQL queries via the warehouse connection.\n
- Sanitization: The instructions provide specific guidance on identifier quoting and column casing for BigQuery, Snowflake, and Databricks to ensure data is correctly and safely interpreted by the data warehouse.
Audit Metadata