semantic-kernel

Why we included this project

Semantic Kernel is a model-agnostic SDK for teams that want to ship LLM-powered features without hand-rolling every integration. It gives you a single programming model for connecting to OpenAI, Azure OpenAI, Hugging Face, and local runtimes like Ollama, then composing those connections into agents and multi-agent workflows. Python, .NET, and Java all get first-class bindings, so a polyglot shop can standardize on one abstraction across services. The plugin system lets you wrap native code, prompt templates, OpenAPI specs, or MCP tools as callable functions, which makes it practical for wiring real business logic into an assistant rather than just demoing chat. And since Microsoft is evolving this codebase into its Agent Framework, the API surface is stable and backed by long-term support, which matters when you are putting something into production.