leanmcp-sdk

Production-ready TypeScript SDK for MCP servers: auth, multi-tenant, observability. Build enterprise AI agents fast.

Installation

Run this command in your terminal to add the MCP server to Claude Code.

Run in terminal:

Command

claude mcp add --transport stdio leanmcp-leanmcp-sdk node dist/main.js \
  --env PORT="8080" \
  --env NODE_ENV="production"

How to use

LeanMCP SDK provides production-ready infrastructure for building MCP servers with authentication, multi-tenant isolation, request logging, observability, and deploy-ready tooling. This repository demonstrates a Node.js-based MCP server scaffold that you can customize by defining services, inputs, and outputs with the LeanMCP core decorators. To use it, install the project, install core dependencies, and start the server. The server exposes an HTTP endpoint at /mcp and a health check at /health, enabling you to register and invoke your MCP tools, validate inputs, and orchestrate multi-step tool interactions. You can leverage the included CLI patterns (via @leanmcp/cli) to scaffold new servers, define services, and generate properly structured MCP endpoints for your domain.

How to install

Prerequisites:

Node.js (16+ recommended) and npm/yarn installed
Basic familiarity with TypeScript projects

Install CLI tools globally (optional but recommended):

npm install -g @leanmcp/cli

Create a new MCP server project (example):

npx @leanmcp/cli create my-mcp-server
cd my-mcp-server
npm install

Install core MCP packages in your project (if not already present):

npm install @leanmcp/core

Start the server:

npm start

Verify runtime locally:

HTTP endpoint: http://localhost:8080/mcp
Health: http://localhost:8080/health

Additional notes

Tips:

Port and environment can be overridden with environment variables (e.g., PORT, NODE_ENV).
Define your services under the mcp/ directory following the examples in the Quick Start to ensure proper schema validation and tool registration.
Use the LeanMCP decorators (Tool, Optional, SchemaConstraint) to describe inputs/outputs for your tools clearly.
Ensure your deployment environment exposes the required ports and that observability/monitoring tooling is configured for production deployments.