Newsletter
Join the Community
Subscribe to our newsletter for the latest news and updates
Open-source UI component library for building MCP apps for ChatGPT, Gemini, Claude, and MCP clients like Cursor.

A full-stack Next.js TypeScript template focused on functionality over UI, featuring Zustand, TanStack Query, and ESLint 9 support.
MCP UI by Flowbite is an open-source template and component library for building Model Context Protocol (MCP) apps that run inside ChatGPT, Gemini, Claude, and other MCP clients such as Cursor and Windsurf.
MCP UI by Flowbite is an open-source library of UI components, widgets, and examples for building apps that work as ChatGPT apps, Gemini apps, Claude apps, and any other MCP client such as Cursor or Windsurf. It is built on top of the Flowbite UI framework and the SkyBridge framework, and the repository provides both a backend MCP endpoint and a React frontend for rendering components.
The template runs an Express server on port 3000 that exposes an MCP endpoint at /mcp, alongside a Vite HMR dev server serving the React-based UI elements. It requires Node.js version 24 or later and an HTTP tunnel such as ngrok to expose the local server to MCP clients.
index.css.npm install, then npm run dev to start the Express and Vite servers, then expose with ngrok http 3000 and register the forwarding URL with the /mcp suffix.This template is aimed at developers and teams building MCP apps for AI assistants. It is useful for:
gemini mcp add --transport http and use the discovered tools in the terminal.mcp.json and let the agent call the tools automatically.The repository is structured as a single Node.js project. After installing dependencies with npm install (or bun, pnpm, or yarn), running npm run dev starts an Express server on port 3000 with an MCP endpoint at /mcp, plus a Vite HMR server for the React app. You then create a public tunnel with ngrok http 3000, take the generated forwarding URL, append /mcp, and register it as a connector in your chosen MCP client. The client can then call the available tools, which return Flowbite-styled UI elements.
Yes, the repository is open-source and available on GitHub under the Flowbite organization.
The documentation includes installation guides for ChatGPT, Claude Web, Claude Code, Gemini CLI, Cursor, VS Code, Mistral AI, and Codex. Any MCP client that supports remote HTTP MCP servers can use the same endpoint.
The project requires Node.js 24 or later, as stated in the prerequisites. An HTTP tunnel such as ngrok is also required for exposing the local server during development.
The included examples are basic text, radio group, checkbox group, line chart, e-commerce products, QR code, and table data. These are built with Flowbite UI components.
Yes, theming is handled through index.css using Flowbite’s theming system. You can choose from predefined themes like default, minimal, enterprise, playful, and mono, or create your own.
