MCP Quickstart

MCP Quickstart is an MCP server project and reference implementation that demonstrates how to build and integrate Model Context Protocol components with OpenAI APIs and Claude Desktop. Built for developers learning protocol implementation patterns, it connects language models directly to custom tools, dynamic resources, and full-stack web applications. The repository organizes several working architectures, including a weather MCP server paired with a client routing tool calls via OpenAI Chat Completions and Responses APIs. It also provides a calculator MCP server that exposes protocol resources alongside an interactive HTML UI widget via MCP Apps for Claude Desktop. Additionally, the project features a ChatKit-based full-stack chat application running a FastAPI backend that handles multi-mode messaging, LangChain agent workflows, and live tool-call streaming. Developers use this collection to study end-to-end patterns for executing local tools, managing external API integrations, and rendering agent UI components within supported environments.

Category: Other & General Purpose

Tags: example, openai, starter, template

Visit MCP Quickstart

How to install and configure MCP Quickstart

1. Clone the repository locally: git clone https://github.com/veysby/mcp-quickstart. 2. Ensure your environment meets the prerequisites: Python 3.13+ (or Python 3.14+ for the chatkit backend) with uv, Node.js 18+ with npm (for the frontend), and an OpenAI API key. 3. Navigate to the specific project directory you want to run (weather/, calculator/, or chatkit/). 4. For Claude Desktop integration with the calculator server or custom tool setups, refer directly to the respective subfolder README.md files in the repository for specific client registration details: https://github.com/veysby/mcp-quickstart.

What you can do with MCP Quickstart

• Route weather forecast and alert tool calls through OpenAI Chat Completions and Responses APIs using a custom Python client. - Render an interactive HTML calculator widget directly inside Claude Desktop using MCP Apps and protocol resources. - Stream LangChain agent tool-call execution steps live within a React and Vite web chat application. - Run a multi-mode FastAPI chat backend that dynamically switches between echo handling and tool-backed agent execution.

Key facts

• https://github.com/veysby/mcp-quickstart
• Other & General Purpose
• example, openai, starter, template

Part of MCP Servers

What is MCP Quickstart?

MCP Quickstart is a collection of reference projects showcasing how to build Model Context Protocol servers and integrate them with OpenAI APIs, Claude Desktop, and web interfaces.

What components are included in MCP Quickstart?

The project includes a weather server with a custom OpenAI client, a calculator server featuring MCP resources and an interactive UI widget for Claude Desktop, and a full-stack ChatKit web app backed by FastAPI.

What prerequisites are needed to run MCP Quickstart?

You need Python 3.13 or newer with uv, Node.js 18 or newer with npm for the ChatKit web app, an OpenAI API key, and Claude Desktop if testing interactive MCP Apps widgets.

Which MCP clients work with MCP Quickstart?

It works with custom Python clients using OpenAI APIs, Claude Desktop for the interactive calculator widget, and full-stack web interfaces built using OpenAI ChatKit and FastAPI.