The Agent Skills Directory

Infrastructure Management and Command Execution: The skill automates the deployment of reliability fixes by executing commands such as azd up, terraform apply, and az deployment group create. These actions modify live cloud resources based on user confirmation.
Sensitive Configuration Access: To evaluate storage redundancy, the skill retrieves application settings including AzureWebJobsStorage connection strings. This information is used locally to identify and verify the underlying storage accounts and is not transmitted outside the user's environment.
Indirect Prompt Injection Surface: The skill processes data from Azure resource metadata which could potentially contain instructions intended to influence the agent's behavior. (1) Ingestion points: Azure Resource Graph results and application settings retrieved from the Azure CLI. (2) Boundary markers: The instructions do not specify explicit delimiters for external data. (3) Capability inventory: The skill can execute CLI commands (az, azd, terraform) and modify local IaC files. (4) Sanitization: There is no explicit sanitization of ingested resource metadata before use in configuration generation.
Trusted Resource Dependencies: The skill utilizes official Microsoft tools and Azure Verified Modules (AVM) for infrastructure configuration, such as the resource-graph extension and official Bicep registry modules, which are trusted components of the Azure ecosystem.

azure-reliability