Skip to content

Getting Started with MCP-Builder.ai


MCP-Builder.ai enables you to easily create and deploy your own MCP-Server with customizable configurations, multiple transport protocol options, and flexible deployment methods.

MCP-Builder.ai is using the powerful APICHAP framework. The underlying configuration language of APICHAP is used to build and connect your MCP-Server to other datasources. Check out the configuration pages to learn more about the Syntax.

How to Start Your MCP-Server Built with MCP-Builder.ai

MCP-Builder.ai supports various transport protocols and deployment functionalities to fit different use cases:

Supported Transport Protocols of the Model Context Protocol (MCP)

MCP-Builder.ai is based on the Model Context Protocol, a modern API standard optimized for real-time communication between model servers and clients.

Available Transport Protocols:

  • STDIO:
    Standard input/output communication. Ideal for local development and testing environments.
    ✅ Supported in MCP-Builder.ai

  • SSE (Server-Sent Events):
    One-way communication protocol that streams events from the server to the client. Useful for updating clients with real-time data.
    ✅ Supported in MCP-Builder.ai
    ⚠️ Note: SSE is deprecated and will soon be replaced.

  • HTTP-Streamable:
    Modern, efficient protocol offering bidirectional streaming over HTTP.
    🚫 Not yet supported in MCP-Builder.ai (Coming Soon)

How to Deploy Your MCP-Server

MCP-Builder.ai provides three primary deployment methods depending on your needs:

1. Run Locally Using STDIO

For quick setup and development purposes.

Requirements:

  • Java 17 or higher installed on your system.

How to Run:

java -jar /path/to/your.jar

Depending on your client setup, you may need to configure the JAR file path in the client settings. For Claude Desktop Users:

Refer to the official quickstart guide: https://modelcontextprotocol.io/quickstart/user

2. Containerized On-Premise Installation

For production environments or private infrastructure deployments.

MCP-Builder.ai will ship your server as a ZIP package containing a pre-configured containerized application using Docker Compose or Podman Compose.

Requirements:

  • Docker or Podman installed on your system.

How to Run:

  • Unzip the package.
  • Follow the instructions in the included README.md.
  • Start your MCP-Server using:
  • The provided startup script, or
  • docker compose up / podman compose up.

Currently the container is exported using SSE protocol but will soon be replaced by the new HTTP-Streamable.

3. Run in MCP-Builder.ai Cloud (Coming Soon)

Easiest deployment option—hosted directly in MCP-Builder.ai Cloud.

Simply click "Run," and we handle the hosting of your HTTP-Streamable MCP-Server for you.

How to Use Your MCP-Server with Your Client

Once your MCP-Server is running, connect it to your preferred client.

Learn how to implement the MCP-Server with your client by following the official quickstart guide: https://modelcontextprotocol.io/quickstart/client