Semantic Kernel

framework active open-source

A Microsoft SDK for integrating LLMs into applications with enterprise-grade patterns. Semantic Kernel emphasizes plugins (skills), planners, and memory as first-class concepts, with strong .NET support alongside Python.

Implements

Concepts this tool claims to implement:

  • Planner abstractions (SequentialPlanner, StepwisePlanner) that decompose goals into plugin invocations. Agent loop with planning and execution.

  • Plugin system with native functions and semantic functions (prompt-based). Strong typing and OpenAPI integration.

  • Agent Memory secondary

    Memory abstractions with semantic memory (embeddings), volatile memory, and persistent memory connectors.

  • Semantic functions are prompt templates with input variables. Prompt engineering as a first-class concept.

Integration Surfaces

  • .NET SDK
  • Python SDK
  • Java SDK
  • Azure AI integrations

Details

Vendor
Microsoft
License
MIT
Runs On
local, cloud
Used By
human, agent, system

Notes

Semantic Kernel is Microsoft's enterprise-focused alternative to LangChain. It has the best .NET support in the ecosystem and integrates deeply with Azure AI services. The plugin model is well-suited for enterprise scenarios with existing APIs and services.