microsoft/semantic-kernel
Integrate cutting-edge LLM technology quickly and easily into your apps
What it solves
Semantic Kernel 是一个企业级的编排框架，旨在简化 AI 代理和多代理系统的构建、部署与管理。它提供一个模型无关的 SDK，允许开发者在应用中集成各种大型语言模型（LLM），同时保持灵活性和可靠性。
How it works
它充当 AI 模型与应用逻辑之间的桥梁。开发者可以创建配备插件（原生代码函数、提示模板或 OpenAPI 规范）、通过向量数据库提供记忆以及规划能力的模块化 AI 代理。该框架支持多代理编排，允许专门的代理协同工作以解决复杂工作流。
Who it’s for
它面向构建 AI 驱动应用的开发者，范围从简单的聊天机器人到复杂的企业级多代理系统，支持 Python、.NET 或 Java。
Highlights
- Model Flexibility: Supports OpenAI, Azure OpenAI, Hugging Face, and NVIDIA, as well as local deployments via Ollama, LMStudio, or ONNX.
- Multi-Agent Orchestration: Ability to coordinate multiple specialized agents for complex business processes.
- Plugin Ecosystem: Extensible via native code, prompt templates, and the Model Context Protocol (MCP).
- Enterprise Features: Built-in support for observability, security, and integration with vector databases like Azure AI Search, Elasticsearch, and Chroma.