The Agent Skills Directory

Command Execution: The skill includes instructions for using shell-based management tools like the Azure CLI or Azure PowerShell as a fallback when platform-native tools are unavailable. This is a standard administrative pattern for cloud management. The skill mitigates potential risks by specifically advising the agent not to use shell environments for general data parsing or aggregation, focusing its use strictly on bounded API calls.
Indirect Prompt Injection: The skill ingests data from the Azure environment, including budget names, resource tags, and cost attribution details (identified in references/budget-health.md and references/guardrails.md). As with any tool that processes cloud metadata, there is a potential surface where malicious content stored in these fields could influence the agent's behavior. The skill provides guidance on confirming scopes and values to help manage this risk surface.
Data Protection and Governance: The skill handles sensitive financial and configuration data. It includes explicit security-minded instructions to prevent the exposure of access tokens and ensures that critical actions, such as budget creation, are only performed after explicit user confirmation. These measures align with best practices for managing sensitive enterprise environments.

cost-governance