lofcz/LLMTornado

What it solves

LLM Tornado is a .NET SDK designed to simplify the development of AI agents and workflows by providing a provider-agnostic interface. It eliminates the need to integrate multiple first-party SDKs and allows developers to build applications that can switch between different AI providers (like OpenAI, Anthropic, and Google) simply by changing a model name.

How it works

The SDK provides built-in connectors to over 30 API providers and vector databases. It uses a unified API for inference, orchestration, and multimodal inputs/outputs. For agentic behavior, it implements an orchestration system based on Orchestrator (graphs), Runner (nodes), and Advancer (edges). It also supports cutting-edge protocols like the Model Context Protocol (MCP) for data source connection and Agent2Agent (A2A) for cross-platform collaboration.

Who it’s for

.NET developers building AI-powered applications, ranging from simple chatbots to autonomous agents and enterprise-grade AI workflows.

Highlights

• Provider Agnostic: Built-in connectors for 30+ providers including Alibaba, Anthropic, Azure, DeepSeek, Google, Groq, and OpenAI.
• Local Deployment Support: Integrated support for vLLM, Ollama, and LocalAI.
• Agent Orchestration: Tools for coordinating specialist agents with support for parallel execution and Mermaid export.
• Multimodal Capabilities: Supports text, images, videos, documents, URLs, and audio.
• Vector Database Integration: Connectors for Chroma, PgVector, Pinecone, Faiss, and QDrant.
• Enterprise Features: Includes a guardrails framework, Open Telemetry support, and request transformation capabilities.
• Interoperability: Compatible with Microsoft.Extensions.AI and Semantic Kernel.