Why should you care?

Microsoft’s agent-style apps in .NET commonly use SK to keep tool calling and prompts organized.

See it live — copy this example

Run CLI steps in a terminal. Run C# samples in a .NET 8 console or Web API project with Ollama running on localhost:11434.

// dotnet add package Microsoft.SemanticKernel
using Microsoft.SemanticKernel;

var builder = Kernel.CreateBuilder();
// Next lesson: add Ollama chat completion
var kernel = builder.Build();
Console.WriteLine("Kernel ready");

What happened?

• CreateBuilder starts configuration.
• Build() creates the Kernel object your plugins and chat services hang off.

Practice next

1. Create a Web/console app.
2. Add the SemanticKernel package.
3. Build an empty kernel.
4. Print kernel.Plugins.Count.
5. Add logging to the builder.

Remember

SK organizes LLM + tools in .NET. Kernel is the hub object. Plugins expose tools to the model.

SK as the agent spine

LocalAIDesk uses one Kernel per request scope.

Outcome: Clear place to register tools.