KimiFoundryDemo

A minimal .NET console app that calls a chat model (default: Kimi-K2.7-Code) hosted on Azure AI Foundry using the Azure.AI.Inference SDK with Entra ID (passwordless) authentication.

• .NET 10 SDK
• Azure CLI
• An Azure AI Foundry resource with a deployed model
• The Cognitive Services User (or Azure AI Developer) role assigned to your identity on the Foundry resource

The app depends on the following NuGet packages (already referenced in KimiFoundryDemo.csproj):

To add them to a fresh project:

dotnet add package Azure.AI.Inference --version 1.0.0-beta.5
dotnet add package Azure.Core --version 1.60.0
dotnet add package Azure.Identity --version 1.21.0

To restore the packages for this project:

dotnet restore

The app reads two environment variables:

1. Sign in with the Azure CLI (used by DefaultAzureCredential):

   az login

2. Set the endpoint (and optionally the model) for the current terminal session:

   $env:AZURE_AI_CHAT_ENDPOINT = "https://<your-resource>.services.ai.azure.com/models"
   $env:AZURE_AI_MODEL = "Kimi-K2.7-Code"

3. Run the app:

   dotnet run

You should see the model's code review printed to the console.

• Authentication uses DefaultAzureCredential with ExcludeManagedIdentityCredential = true, so it does not probe the Azure IMDS endpoint (169.254.169.254) when running locally.
• Foundry (Cognitive Services) endpoints require a token whose audience is https://cognitiveservices.azure.com. The app wraps the credential (ScopeOverrideCredential) to force this scope.