cost-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection: The skill is designed to ingest and analyze data from Azure Cost Management and Resource Graph APIs. This ingestion of external data is central to its purpose but is managed through specific validation steps. \n
- Ingestion points: Data enters the context via the
query_costs,query_aks_costs, andexecute_querytools as described inSKILL.mdand associated workflow references. \n - Boundary markers: The instructions emphasize labeling incomplete data and separating currency results, although they do not define specific text delimiters for the interpolation of API responses. \n
- Capability inventory: The skill utilizes read-only cost query tools and allows for administrative actions (like enabling AKS cost analysis) via approved fallback CLI tools. \n
- Sanitization: Safety is maintained through a mandatory
validate_querystep for Resource Graph and a requirement for explicit user confirmation before any configuration changes or the use of fallback shell commands.\n- Command Execution: The skill provides documentation for falling back to Azure CLI or PowerShell commands if specific tool functionality is missing. This is a common pattern for cloud management skills and is mitigated by instructions that strictly require user approval and prohibit using these tools solely for parsing data.
Audit Metadata