Semantic Kernel
used in
0
stacks
Microsoft's open-source SDK for embedding LLM agents and plugins into .NET, Python, and Java apps.
Provides a kernel abstraction with plugins, connectors, memory, and agent primitives for orchestrating LLM calls inside enterprise applications, with first-class .NET, Python, and Java support and Azure integration. It remains supported, but Microsoft has consolidated Semantic Kernel and AutoGen into the new Microsoft Agent Framework (1.0 released April 2026), where most new investment now goes. Existing projects have a documented migration path to that successor.
Strengths
- Strong enterprise and multi-language support (.NET, Python, Java)
- Plugin/connector model and Azure ecosystem integration
- Backed by Microsoft with extensive documentation
- Established memory and planner/agent primitives
Tradeoffs
- Most new investment has shifted to Microsoft Agent Framework
- API churn and migration overhead for long-lived projects
- Heavier, enterprise-oriented abstractions