cost-optimization
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- Indirect Prompt Injection Surface: The skill ingests data from external Azure sources, including Azure Advisor recommendations and resource tags. There is a security consideration that untrusted text within these fields could be incorporated into the agent's output. The skill addresses this by using structured report templates and focusing on factual inventory data.
- Command-Line Tool Fallback: The documentation allows for the use of authenticated Azure CLI or PowerShell commands if primary operations are unavailable. This pattern is restricted to official administrative tools and explicitly prohibits the use of local interpreters for data parsing, reducing the risk of unauthorized execution.
- Dynamic Query Execution: The skill generates and executes KQL queries through Azure Resource Graph. This dynamic execution is mitigated by a required validation step and instructions to bound result counts, ensuring that resource analysis remains within expected operational parameters.
Audit Metadata