The Agent Skills Directory

Official Repository Integration: The skill facilitates the cloning of the reference application directly from the official Microsoft GitHub organization. This is a standard method for distributing sample code and ensures the source is verified.
Infrastructure Provisioning Script: A provided shell script, provision-azure-openai.sh, automates the setup of Azure OpenAI resources using standard Azure CLI commands. This script is intended for local execution to configure the user's development environment.
Standard Secret Management: The instructions guide users to store sensitive credentials, such as API keys, in local .env files within the application directories. This aligns with industry-standard security practices for keeping secrets out of source control and logs.
Trusted Service Connectivity: The skill performs connectivity tests and API requests to well-known services, specifically Azure OpenAI and OpenAI endpoints, to verify configuration during the setup process.
Prompt Processing Considerations: The skill's primary function involves ingesting user prompts to generate dynamic UI controls.
Data Ingestion Surface: Untrusted user input enters the system through prompt fields processed by lib/generate-controls.ts.
Boundary Markers: The documentation recommends using distinct system prompts to separate internal instructions from user-provided data.
Capability Inventory: The skill utilizes model invocation for generation and provides shell commands for environment setup and project building.
Sanitization Measures: The skill includes a dedicated validation module, lib/validate-controls.ts, which is required to check model-generated control schemas for correctness and safety before they are rendered in the UI.

msresearch-promptions