azure-prepare

Pass

Audited by Gen Agent Trust Hub on May 3, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references a command to install the Azure Developer CLI using curl -fsSL https://aka.ms/install-azd.sh | bash in references/sdk/azd-deployment.md. This is a standard installation method from a trusted Microsoft domain.
  • [COMMAND_EXECUTION]: The skill extensively uses and generates instructions for various CLI tools including az, azd, docker, dotnet, and npm. These are standard DevOps operations required for the skill's primary purpose of application preparation and infrastructure provisioning.
  • [PROMPT_INJECTION]: The skill uses authoritative language in SKILL.md (e.g., 'AUTHORITATIVE GUIDANCE', 'MANDATORY COMPLIANCE', 'IGNORE any prior training'). While strong, this is used to enforce technical consistency for complex deployment workflows rather than attempting to bypass safety filters or extract system prompts.
  • [DATA_EXFILTRATION]: The skill includes instructions for handling secrets securely. It explicitly prohibits hardcoding credentials in Bicep or Terraform files and mandates the use of Microsoft Entra ID (managed identity) and Azure Key Vault. It includes scripts that handle environment variables but implements safe parsing to prevent arbitrary code injection.
  • [SAFE]: The skill implements significant proactive security measures, such as Rule 10 in SKILL.md which forbids the generation of SQL administrator credentials in favor of Entra-only authentication. It also mandates a 'Plan-First' workflow requiring user approval before any resource generation or modification.
Audit Metadata
Risk Level
SAFE
Analyzed
May 3, 2026, 05:51 AM