Which LLMs does Semantic Kernel Agent support?

It features multi-provider LLM routing, allowing seamless switching between Azure OpenAI, OpenAI, Ollama, Google Gemini, and HuggingFace without any code changes.

What is the primary function of Semantic Kernel Agent?

It's an AI agent built on Microsoft Semantic Kernel that automates the conversion of Figma designs into JSX-ready code by extracting design tokens like text, colors, and layout.

How does it extract design information from Figma?

It integrates with Figma's API via a custom Model Context Protocol (MCP) server, which extracts design tokens and provides this structured data for the agent to convert into JSX.

Semantic Kernel Agent

Name: Semantic Kernel Agent
Author: AnithaKarre

0•

Transforms Figma designs into JSX-ready code by extracting design tokens and leveraging multi-provider LLM routing and modular prompt templates.

This robust framework empowers AI agents to transform Figma designs directly into production-ready JSX code. Leveraging Microsoft Semantic Kernel and the Model Context Protocol (MCP), it intelligently extracts design tokens like text, colors, and layout details from Figma files, then orchestrates generation using a flexible LLM factory that supports leading providers such as Azure OpenAI, OpenAI, Ollama, Google Gemini, and HuggingFace, ensuring adaptable and efficient design-to-code workflows.

主要功能

01Figma MCP Integration for design token extraction (text, colors, layout) and JSX conversion.

02Multi-Provider LLM Routing to swap between Azure OpenAI, OpenAI, Ollama, Gemini, and HuggingFace without code changes.

03Modular Prompt Factory supporting .txt and .docx templates with layered composition (Base → Business → Knowledge).

040 GitHub stars

05FastAPI-based Multi-Agent Orchestration with custom agent selection and session isolation.

06Stateless transport for MCP server using SSE over HTTP, compatible with proxies and load balancers.

使用案例

01Implementing flexible LLM-powered design-to-code pipelines with support for multiple AI providers.

02Building AI agents that can interact with design tools like Figma for token extraction.

03Automating the conversion of Figma design mockups into JSX code components.