create-wi
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides automation scripts (Bash and PowerShell) that interface with the Azure CLI and Azure DevOps REST APIs. These scripts perform administrative tasks including resource group creation, managed identity management, and service connection configuration.
- [PROMPT_INJECTION]: The skill processes untrusted input from a local JSON configuration file which presents an indirect prompt injection surface. Ingestion points: Metadata is read from .github/deploy.json in the repository root. Boundary markers: None are present to delimit or sanitize the injected metadata. Capability inventory: The skill executes commands with high privileges, including assigning the Contributor RBAC role and creating federated credentials in Azure. Sanitization: There is no evidence of validation or escaping for the data read from the configuration file before its use in command execution.
Audit Metadata