The Agent Skills Directory

[Command Execution]: The skill is designed to interact with the Azure CLI (az) to perform tasks such as listing model capabilities, checking subscription quotas, and creating deployments. These actions use parameters like subscription IDs and location strings derived from user input, which is the standard mechanism for the skill's resource management functionality.
[Deployment Confirmation]: A specific safety protocol is included that requires the agent to display the resolved project, region, and resource group to the user for explicit confirmation before any deployment action is taken, reducing the risk of accidental or incorrect resource provisioning.
[Indirect Prompt Injection Surface]: The skill processes user prompts to determine deployment intent and parameters, representing an ingestion point for external data.
Ingestion points: User-provided deployment requests and parameters in the prompt (as described in SKILL.md).
Boundary markers: The skill mandates a confirmation step and dynamic validation of inputs against the Azure model catalog.
Capability inventory: Execution of az CLI commands for deployment management and quota verification.
Sanitization: Relies on dynamic querying of the Azure API via CLI to validate user-supplied parameters like model names and versions.
[Resource URL Construction]: The skill includes helper scripts to generate links to the Microsoft Foundry portal (ai.azure.com). These scripts perform data formatting, such as Base64URL encoding of the subscription ID, to create valid URLs for the vendor's web interface.

deploy-model