monte-carlo-storage-cost-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes and displays external data retrieved from data warehouses (Snowflake, BigQuery, Redshift, Databricks) via the
analyze_storage_costsMCP tool, creating a surface where malicious metadata in the warehouse could influence agent behavior. - Ingestion points: Tool output from
analyze_storage_costsis interpolated into the agent context inSKILL.md(Step 3 and Step 4). - Boundary markers: The skill utilizes specific HTML-style boundary markers (
<!-- PRESENT_AS_IS -->,<!-- CATEGORY_DETAILS -->, and<!-- CATEGORY:<key> -->) to help the agent distinguish between summary data and detailed drill-down data. - Capability inventory: The skill is restricted to read-only analytical tools (
analyze_storage_costs,get_asset_lineage) and explicitly states that destructive actions likeDROP TABLEare out of scope. - Sanitization: The instructions direct the agent to present tool output verbatim without explicit sanitization or escaping of the content provided by the warehouse metadata.
Audit Metadata