Skip to content
Get packs

Semantic Kernel

Open-source SDK for building AI agents and multi-agent apps in Python, .NET and Java.

GitHub Download dotnet-1.81.0
Stars
28.7k
Forks
4.9k
Licence
MIT
Language
C#
Last update
4 days ago
Type
Library

What it is

Semantic Kernel is a model-agnostic SDK from Microsoft that lets developers connect large language models (such as OpenAI or Azure OpenAI models) to their own apps, add tools and plugins, and orchestrate one or more AI agents. The README says it is now succeeded by Microsoft Agent Framework, so new projects should check that project first.

  • Build a chat assistant with custom instructions using ChatCompletionAgent
  • Give an agent plugin functions, such as a menu lookup, that it can call
  • Request structured output in a defined format from a model
  • Connect an app to vector stores such as Azure AI Search, Elasticsearch or Chroma
  • Run models locally with Ollama, LMStudio or ONNX

Great for

  • Developers already using Azure OpenAI or OpenAI who want a structured SDK for agents
  • Teams working in .NET, Python or Java who want one framework across languages
  • Projects that need plugins built from native code, prompt templates, OpenAPI specs or MCP
  • Existing Semantic Kernel users maintaining or migrating older code

Not ideal for

  • Non-programmers: it is a developer SDK and needs code to use, so no-code users should look elsewhere
  • Teams starting fresh who want the newest path: the README directs users to Microsoft Agent Framework as the successor
  • Simple scripts that only need one direct API call, where an SDK adds more structure than needed
Step-by-step guide: how to use Semantic KernelInstall, first run, examples, tips and FAQs
Semantic Kernel alternativesLangChain vs Semantic KernelAI Agents for Beginners vs Semantic KernelOpenAI Cookbook vs Semantic KernelCrewAI vs Semantic KernelSemantic Kernel vs spaCyLlamaIndex vs Semantic KernelMicrosoft Agent Framework vs Semantic KernelCopilot SDK vs Semantic Kernel

How to get it

  1. 1

    Install with a package manager

    You'll need: Python 3.10+ or .NET 10.0+ or JDK 17+ · An Azure OpenAI or OpenAI API key

    Set Azure OpenAI API key

    export AZURE_OPENAI_API_KEY=AAA....

    Set OpenAI API key (alternative)

    export OPENAI_API_KEY=sk-...

    Install Python package

    pip install semantic-kernel

    Install .NET packages

    dotnet add package Microsoft.SemanticKernel
  2. 2

    Run it

    • Install the package for your language using the commands above
    • Set your Azure OpenAI or OpenAI API key as an environment variable
    • Copy the Basic Agent quickstart for Python or .NET from the README
    • Run it to get a short haiku response from the agent
  3. 3

    Get the most out of it

    • Add plugins with the kernel_function decorator (Python) or KernelFunction attribute (.NET) to let agents call your code
    • Use structured output settings to get replies in a defined format
    • Connect vector database integrations such as Azure AI Search or Chroma for retrieval
    • Try the Multi-Agent System example in the README to coordinate specialist agents

Downloads

Source code (.zip)main branch · straight from GitHubRelease dotnet-1.81.0Release notes and sourceOfficial website & docsaka.ms/semantic-kernel

Strengths

  • Supports several models and providers, including OpenAI, Azure OpenAI, Hugging Face and NVidia, per the README
  • Offers examples in Python and C# with matching quickstarts
  • Permissive MIT license and a large public project with about 28,700 stars

Trade-offs

  • The README states the project is succeeded by Microsoft Agent Framework, so long-term direction may point elsewhere
  • Quickstarts assume Azure or OpenAI credentials and a configured environment, so setup needs some engineering work
  • Java support is pointed to a separate build guide rather than documented here

How it compares

ProjectStarsTypeLevelUpdatedBest for
semantic-kernel (this)28.7kLibraryintermediate4 days agoDevelopers already using Azure OpenAI or OpenAI who want a structured SDK for agents
langchain149kFrameworkintermediate4 days agoPython developers building LLM-powered applications
llama_index52.8kLibraryintermediate10 days agoPython developers building retrieval-augmented generation (RAG) apps over private documents
agent-framework14.5kLibraryintermediate4 days agoDevelopers building agents or multi-agent workflows they plan to run in production
OpenHands92.5kAppintermediate4 days agoDevelopers who want coding agents running on a server while their laptop is off
SuperAGI17.7kAI model / agentadvanced1.8 years agoPython developers experimenting with autonomous AI agents

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.