azure-kusto-irql

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [Command Execution]: The skill contains procedures to generate and execute PowerShell scripts on the host system.
  • Evidence: The file references/KUSTO_EXPLORER_LAUNCH.md provides instructions for the agent to use Set-Content and Start-Process to save .kql files and launch the Kusto Explorer application.
  • Context: This is designed as a convenience feature to automate the transition from a generated query to a local analysis tool. While automated command generation is a security consideration, the skill includes explicit developer guidance to avoid unsafe shell interpolation (such as PowerShell here-strings) to mitigate command injection risks.
  • [Data Exposure Consideration]: The skill facilitates access to schema-unified views of security events that may contain sensitive information.
  • Evidence: The Get_Event_Authentication_All function projects columns including PasswordHash, as noted in SKILL.md.
  • Context: Accessing authentication metadata is a standard requirement for cybersecurity threat hunting (e.g., detecting pass-the-hash attacks). Users should ensure that access to these functions is governed by appropriate role-based access controls within the target Kusto environment.
  • [Indirect Prompt Injection Surface]: The skill ingests user-provided indicators to generate hunting queries, which presents a potential injection surface.
  • Ingestion points: User-supplied indicators or search terms are interpolated into KQL query templates (e.g., Filename has "<INDICATOR>" in SKILL.md).
  • Boundary markers: The skill uses standard KQL let statements and quoted strings as delimiters, but does not explicitly instruct the agent to sanitize all possible KQL injection characters in user input.
  • Capability inventory: The skill utilizes tools such as kusto_query to execute the generated code and PowerShell for file system interaction.
  • Sanitization: The documentation in references/KUSTO_EXPLORER_LAUNCH.md proactively recommends using Set-Content rather than here-strings to prevent command injection during the file creation process.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 06:49 AM
Security Audit — agent-trust-hub — azure-kusto-irql