Semantic Kernel is Microsoft's agent framework for .NET, Python, and Java. It abstracts LLM providers, plugin authoring (with auto-function-calling), planning (Handlebars, Stepwise), memory connectors, and Azure-native auth — designed for enterprise apps where governance and integration matter as much as model quality.
Build → register → invoke flow with auto function calling.
SK ships with Azure auth (DefaultAzureCredential), telemetry (OTel), filters, and content safety integrations.
[KernelFunction] methods auto-export as OpenAI function tools. No manual JSON schemas to maintain.
Author prompts as YAML+text files separate from code. Reviewable, versionable, hot-reloadable.
For multi-step tasks, FunctionCallingStepwisePlanner produces explicit plans you can show + approve before execution.
Use IPromptRenderFilter / IFunctionInvocationFilter for cross-cutting concerns: logging, redaction, caching.
Same concepts across .NET, Python, Java. Keep your team aligned even with mixed stacks.
Sign in to share your feedback and join the discussion.