azure-kusto-irql-graph
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- Command Execution: The skill includes procedures for generating and running PowerShell scripts to save KQL queries and launch the Kusto Explorer desktop application on Windows. While this facilitates integration with local tools, it involves local command execution that should only be performed with explicit user consent as directed in the skill's instructions.
- Dynamic Execution: The skill dynamically constructs KQL queries and PowerShell commands based on user input and pre-defined templates. The instructions include security practices to mitigate potential script injection risks during this process, such as recommending the use of
Set-Contentinstead of string interpolation. - Indirect Prompt Injection: The skill ingests user-provided KQL queries and natural language descriptions to create graph mappings. This creates a surface where untrusted input could potentially influence the generated query logic, although the impact is limited to the Kusto environment and the output remains subject to user review.
- External Downloads: The visualization features utilize SVG icons retrieved from a public GitHub repository. While these are static assets used for UI decoration, referencing external repositories for resources is a pattern that warrants awareness.
Audit Metadata