NOTE

Microsoft Semantic Kernel

authorgemini-cli aliasessemantic-kernel, ms-kernel-memory, sk-framework titleMicrosoft Semantic Kernel statusactive date2026-04-25 typepermanent

Microsoft Semantic Kernel

Semantic Kernel (SK) is an open-source SDK that acts as the "Middleware" or "Brain" for AI applications. It orchestrates the flow between Large Language Models (LLMs) and conventional code (C#, python, Java).

Core Architecture

  • Kernel: The central hub configured via KernelBuilder. It handles dependency injection and manages the lifecycle of AI services.
  • Plugins: Encapsulated functions (native or semantic) that extend the agent's capabilities.
  • Function Calling: Supports automatic invocation of plugins based on LLM intent.

Kernel Memory (KM)

A specialized service for Retrieval-Augmented Generation (RAG).

  • Ingestion Pipeline: Automates PDF/doc parsing, chunking, and embedding.
  • Proactive Memory: Background tasks that summarize and enrich data during ingestion.
  • Storage: Supports cloud (Azure AI Search) and local (Qdrant, Disk, In-Memory) vector stores.

Implementation Pattern (.NET)

var builder = Kernel.CreateBuilder();
builder.AddAzureOpenAIChatCompletion(...);
builder.Plugins.AddFromType<MyPlugin>();
var kernel = builder.Build();

// Automatic function calling
var result = await kernel.InvokePromptAsync("What is the current vault status?");

References