@agentailor/create-mcp-server

Scaffold production-ready MCP servers in seconds.

Quick Start

Interactive mode (guided prompts):

npx @agentailor/create-mcp-server

CLI mode (all options via arguments):

npx @agentailor/create-mcp-server --name=my-server

CLI Options

Option	Short	Default	Description
--name	-n	—	Project name (required in CLI mode)
--package-manager	-p	npm	Package manager: npm, pnpm, yarn
--framework	-f	sdk	Accepted for compatibility; sdk is the only framework
--stdio	—	false	Use stdio transport (for local clients)
--template	-t	—	Deprecated, no effect; prints a warning and will be removed in a future release
--oauth	—	false	Enable OAuth (HTTP only, incompatible with --stdio)
--no-git	—	false	Skip git initialization
--no-skills	—	false	Skip adding the tool-design skill
--help	-h	—	Show help
--version	-V	—	Show version

Examples:

# Minimal - uses all defaults (HTTP streamable)
npx @agentailor/create-mcp-server --name=my-server

# stdio server (for local clients)
npx @agentailor/create-mcp-server --name=my-server --stdio

# Full HTTP options
npx @agentailor/create-mcp-server \
  --name=my-auth-server \
  --package-manager=pnpm \
  --oauth

# Short flags
npx @agentailor/create-mcp-server -n my-server -p yarn --stdio

Features

• MCP SDK v2 — projects serve protocol revision 2026-07-28 and still accept 2025-era clients
• Two transport types — HTTP (streamable) or stdio (for local clients)
• Stateless by design — the SDK handler builds a fresh server per request, so HTTP servers scale without sticky sessions
• Optional OAuth — OIDC-compliant authentication (HTTP only) (setup guide)
• Package manager choice — npm, pnpm, or yarn
• TypeScript ready — ready to customize
• Docker ready — production Dockerfile included (HTTP transport)
• MCP Inspector — built-in debugging with npm run inspect

Official MCP SDK

Projects are generated against the MCP TypeScript SDK v2 — the split @modelcontextprotocol/server, @modelcontextprotocol/express, and @modelcontextprotocol/node packages, rather than the v1 @modelcontextprotocol/sdk monolith.

HTTP servers are built on createMcpHandler, which runs the server factory once per request. They serve protocol revision 2026-07-28 and also accept 2025-era clients. stdio servers use serveStdio, which negotiates the era per connection.

FastMCP was removed in 0.10.0. It never moved to SDK v2, so FastMCP projects stayed on the 2025-era protocol and shipped without tests. To scaffold one anyway, pin the last release that supports it: npx @agentailor/create-mcp-server@0.9 --framework=fastmcp.

Transport Types

Feature	HTTP (Streamable HTTP)	stdio
Use case	Remote access, cloud deployment	Local clients (Claude Desktop)
Protocol	HTTP	stdin/stdout
OAuth support	✓	—
Docker deployment	✓	—
Port configuration	✓	—

HTTP: Deploy as an HTTP server accessible remotely.

stdio: Run as a local process. Communicates over stdin/stdout. Ideal for local clients. No HTTP server, no port, no Dockerfile generated.

Server Modes (HTTP only)

SDK v2 serves every HTTP request through a single per-request idiom: createMcpHandler builds a fresh McpServer for each call, and 2025-era clients are served through the same path. The stateless/stateful distinction that v1 required no longer changes the generated project.

--template is deprecated: both values produce the same output and the CLI prints a warning when it is passed. It will be removed in a future release, so drop it from scripts. If you need session-based serving with SSE resumability, the SDK still offers NodeStreamableHTTPServerTransport — see the SDK v2 sessions guide.

Generated Project

my-mcp-server/
├── src/
│   ├── server.ts             # Creates the server, registers the primitives
│   ├── tools.ts              # Tool definitions
│   ├── prompts.ts            # Prompt definitions
│   ├── resources.ts          # Resource definitions
│   ├── notes-store.ts        # Example data layer (no MCP imports)
│   ├── notes-store.test.ts   # Store unit tests
│   ├── server.test.ts        # Tests driven through a real MCP client
│   ├── index.ts              # Express app and transport setup
│   └── auth.ts               # OAuth middleware (if enabled)
├── Dockerfile        # Production-ready Docker build
├── package.json
├── tsconfig.json
├── .gitignore
├── .env.example
├── AGENTS.md         # Working notes for humans and coding agents
└── README.md

Projects come with a worked example — a small notes server — and a test suite that runs green immediately, so there is a working pattern to copy when you add your own tools.

Projects also get Agentailor's tool-design skill at .agents/skills/, so a coding agent working in the project has the guidance on hand. It is bundled with the CLI, so this needs no network and adds nothing to scaffolding time. Pass --no-skills to skip it, and run npx skills update -p -y inside the project if you want a newer version.

Every project also gets an AGENTS.md covering the things that are easy to get wrong: how to write a tool an agent can use, how to organize tools as they grow, and the transport-specific traps (state is per-request over HTTP; stdout belongs to the protocol over stdio).

Scripts:

• npm run dev — build and start the server
• npm test — run the test suite
• npm run inspect — open MCP Inspector (update URL in package.json if needed)

Learning Resources

Guide	Description
Create Your First MCP Server in 5 Minutes	Build your first production-ready MCP server. A complete beginner guide to scaffolding a Fetch MCP server with TypeScript.
Securing MCP Servers with Keycloak	Learn how to secure your MCP servers with OAuth authentication using Keycloak.
OAuth for MCP Clients (Next.js + LangGraph.js)	Implement OAuth authentication in your MCP client using Next.js and the MCP SDK.

Need help building MCP servers or agent infrastructure?

I help teams design and ship production-ready AI agent systems (MCP, LangGraph, RAG, memory, performance).

If you’re building something serious on top of this:

→ DM me on LinkedIn

Happy to jump on a short call.

What is MCP?

The Model Context Protocol (MCP) is an open protocol that enables AI assistants to interact with external tools, data sources, and services.

Built by Agentailor

Built by Agentailor — your launchpad for production-ready MCP servers and scalable AI agents. We provide the tools, templates, and expertise to ship AI-powered applications faster.