agents-sdk-provision
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Command Execution]: The skill frequently uses the
azCLI to perform administrative tasks like creating identities, managing app registrations, and provisioning bot services. These actions are within the skill's intended scope but require appropriate user permissions and oversight. - [Indirect Prompt Injection Surface]: The skill processes user-supplied strings for parameters like
APP_NAMEandRESOURCE_GROUPand interpolates them into shell commands. This creates a potential surface for indirect prompt injection. - Ingestion points: User-provided variables in
SKILL.mdused for resource naming and identification. - Boundary markers: The skill does not utilize specific delimiters to isolate user input within the agent's prompts.
- Capability inventory: Includes shell command execution through the Azure CLI and the generation of local configuration files.
- Sanitization: The instructions do not specify explicit validation or sanitization routines for the user input before it is used in command strings.
- [External Downloads]: The skill provides instructions for installing the Bicep CLI (
az bicep install). This is a standard procedure for managing Azure infrastructure as code and targets official Microsoft distribution channels. - [Credential Management]: The skill manages the lifecycle of Azure client secrets, including their generation and rotation. It promotes security best practices by advising users to use environment-based secret management or Azure Key Vault rather than hardcoding credentials in source code.
Audit Metadata