azure-env-builder
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied requirements to generate Azure architecture designs and corresponding deployment scripts. This creates a surface where unvalidated user input could be translated into infrastructure code. The analysis did not find explicit input sanitization or boundary markers between user instructions and code templates.
- Ingestion points: User input provided via the skill's primary argument hint in SKILL.md.
- Capability inventory: PowerShell (pwsh) execution, Azure CLI (az) interaction, and file writing capabilities.
- Boundary markers: Absent in the provided templates.
- Sanitization: Not explicitly implemented in the scaffolding logic.
- [DYNAMIC_EXECUTION]: The skill uses templates to dynamically generate PowerShell deployment scripts (deploy.ps1) and Bicep infrastructure files at runtime. These generated files are intended to be executed to manage Azure resources.
- Evidence: scripts/scaffold_environment.ps1 creates directory structures and populates scripts using string interpolation based on user-provided environment parameters.
- [COMMAND_EXECUTION]: The skill facilitates the execution of system commands for infrastructure validation and deployment.
- Evidence: scripts/validate_bicep.ps1 and scripts/preview_cli.ps1 invoke the az CLI and pwsh to run what-if analysis and other deployment tasks.
- [EXTERNAL_DOWNLOADS]: The skill's VM initialization templates include commands to fetch configuration and installers from external sources.
- Evidence: Fetches the Azure CLI installer from Microsoft's official aka.ms domain and Docker's GPG key from its official repository. These references target well-known and trusted service providers.
Audit Metadata