Skip to main content

​
Getting Started with LeanMCP

Build your first MCP server using our web platform - no coding experience required.

​
What You’ll Build

A simple MCP server that helps AI agents work with your tools. Think of it as creating a UI that AI can understand and use.

​
Step 1: Access the Platform

Go to https://ship.leanmcp.com and login with your Google account to get started.
If you don’t have a Google account, you can still login using your email at https://ship.leanmcp.com/email-login

​
Step 2: Create Your MCP Server

  1. Once logged in, select “MCP Server Generation”
  2. Type in the kind of MCP server you need in the input field
  3. Describe what tools and functionality you want your MCP to provide
MCP Builder Interface The AI will start interacting with you, editing and creating files based on your requirements. This is a collaborative process where you can refine and adjust your MCP server as it’s being built.

​
Step 3: Test Your Server

Before deploying, it’s crucial to test your MCP server:
  1. Click the “Test” button in the interface
  2. This will test your build in a sandbox environment
  3. Any errors or issues will be displayed immediately
  4. Fix any problems by providing additional prompts to the AI
Testing in the sandbox only validates the build process. For end-to-end MCP functionality testing, you’ll need to deploy your server.

​
Step 4: Deploy Your Server

When you’re satisfied with your MCP server:
  1. Click the “Deploy” button
  2. Your deployment will be triggered and processed
  3. Monitor your deployment at /my-deployments
  4. View both deployment logs and runtime logs
If your deployment fails, you can:
  • Review the error logs in /my-deployments
  • Prompt the AI again with more specific requirements
  • Make adjustments and redeploy

​
Alternative: GitHub Integration

If you prefer to build your MCP server directly in your own repository and use our CI/CD pipeline:

​
Install GitHub App

  1. Install our GitHub app at https://github.com/apps/leanmcp-com
  2. We recommend giving permissions only to repositories where you need CI/CD pipelines
  3. This provides better security and cleaner access management

​
Repository Setup

Once the GitHub app is installed:
  • Push your MCP server code to your repository
  • Our platform will automatically detect changes and trigger deployments
  • Monitor deployments through the same /my-deployments interface

​
Troubleshooting

​
GitHub Push Failures

Problem: Pushing code to GitHub repository fails Solutions:
  • Ensure the GitHub repository exists and you have write access
  • Make sure you have already created the main branch
  • Verify the repository URL is correct
Currently, we only support updates to the main branch. Support for other branches is coming soon.

​
Deployment Issues

Problem: Deployment fails or shows errors Solutions:
  • Check the deployment logs in /my-deployments for specific error messages
  • Review your MCP server configuration for any missing requirements
  • Re-prompt the AI with more detailed specifications
  • Ensure all required dependencies are properly defined

​
Testing Problems

Problem: Sandbox testing shows build errors Solutions:
  • Review the error messages carefully - they usually indicate missing dependencies or configuration issues
  • Provide more specific requirements to the AI about your tools and their dependencies
  • Check that all file paths and imports are correct

​
Next Steps

View Your Deployments

Monitor and manage your deployed MCP servers

Learn MCP Concepts

Understand tools, resources, and prompts

Best Practices

Build MCPs that AI agents love to use

GitHub Integration

Connect your repositories for automated deployments

​
Alternative: Command Line Interface

For developers who prefer command-line tools, use the LeanMCP CLI to manage projects locally:

​
Quick CLI Setup

# Install via Homebrew (macOS)
brew tap rosaboyle/mcli
brew install mcli

# Authenticate
mcli auth login --api-key <your-api-key>

# Create project from local directory
mcli projects create --name "My MCP" --path ./

Full CLI Documentation

Complete guide to the LeanMCP CLI

​
Need Help?

Be specific about your tools and use cases. For example: “I want an MCP server that can read and write files, execute shell commands, and manage a todo list stored in JSON format.”
Try breaking down your requirements into smaller, more specific pieces. You can always refine and add more functionality through additional prompts.
Yes! The AI-generated code is fully editable. You can make changes and redeploy as needed.
Most deployments complete within 2-3 minutes. Complex MCP servers with many dependencies may take longer.