MCP-Builder.ai

Connect the MCP-Server with your AI applications!

Step-by-step guides to integrate your MCP-Server with various AI applications.

Transport Protocol: SSE

4 snippets available

OpenAI Playground (SSE)

SSE

Connect your SSE MCP-Server to OpenAI using the playground.

Requirements:

Your containerized server is installed on your server and is using a valid SSL certificate.

1

Start your MCP-Server (Only necessary for On-Premise Setup)

a. Unzip the downloaded zip file.

b. Use the shipped docker-compose.yaml file or create one yourself.

c. Open the docker-compose.yaml file and check for any environment variables that you might need to replace

d. Start your container using the included start.bat or start.sh files or start the containerized compose yourself using: 

# Run on windows: 
start.bat
# Run on linux: 
start.sh
# Run with docker-compose: 
docker-compose up
# Run with podman-compose: 
podman-compose up

e. Verify the logs to see if your MCP-Server started successfully

2

Open platform.openai.com

Open https://platform.openai.com/chat and click "Create"

OpenAI Prompt Setup
3

Add MCP-Server

Click on Tools: "Add"

Add MCP-Server

Add the URL to your MCP-Server using: "https://www.example.com/sse". The /sse must be added.

Add credentials

If your MCP-Server is secured with authentification, then set the credentials under Authentification.

When everything works, you will see the tools available in your MCP-Server.

MCP-Server is connected
3

Add a system message and prompt

No add the wanted system message and test your MCP-Server connection

Test with prompts

OpenAI SDK (SSE)

SSE

Connect your SSE MCP-Server using the OpenAI SDK

Requirements:

Your application must use the OpenAI SDK

The server or local machine has is able to run containerized apps with podman or docker.

1

Start your MCP-Server (Only necessary for On-Premise Setup)

a. Unzip the downloaded zip file.

b. Use the shipped docker-compose.yaml file or create one yourself.

c. Open the docker-compose.yaml file and check for any environment variables that you might need to replace

d. Start your container using the included start.bat or start.sh files or start the containerized compose yourself using: 

# Run on windows: 
start.bat
# Run on linux: 
start.sh
# Run with docker-compose: 
docker-compose up
# Run with podman-compose: 
podman-compose up

e. Verify the logs to see if your MCP-Server started successfully

2

Add your MCP-Server

Add your MCP-Server to your OpenAI SDK code:

from openai import OpenAI

client = OpenAI()

resp = client.responses.create(
    model="gpt-4.1",
    tools=[{
        "type": "mcp",
        "server_label": "your_project_name",
        "server_url": "https://www.linktoyourserver.com/mcp",
    }]
)

MCP-Inspector (SSE)

SSE

Debug your SSE MCP-Server

Requirements:

npx is installed NPX

The server or local machine has is able to run containerized apps with podman or docker.

1

Start your MCP-Server (Only necessary for On-Premise Setup)

a. Unzip the downloaded zip file.

b. Use the shipped docker-compose.yaml file or create one yourself.

c. Open the docker-compose.yaml file and check for any environment variables that you might need to replace

d. Start your container using the included start.bat or start.sh files or start the containerized compose yourself using: 

# Run on windows: 
start.bat
# Run on linux: 
start.sh
# Run with docker-compose: 
docker-compose up
# Run with podman-compose: 
podman-compose up

e. Verify the logs to see if your MCP-Server started successfully

2

Start & Open MCP Inspector

Open the terminal and run:

npx @modelcontextprotocol/inspector

Get more infos at:MCP-Inspector

Run MCP-Inspector
4

Connect with your SSE Server

Enter the url to your SSE Server

If your SSE Server requires any additional header parameters or secrets you have to add them under Authentication.

MCP-Inspector will send all Header Parameters in lower-case. MCP-Builder.ai is handling input parameters case-sensitive - meaning you must make PARAM(xx) values in your specs operation configuration files´s lower case as well.

MCP Inspector Dashboard

Others (SSE)

SSE

Connect your SSE MCP-Server to other AI applications

Requirements:

Your AI application or SDK includes MCP-client functionality and support the SSE Transport Protocol.
See a full list of supported clients here: See which tools support MCP-Client

The server or local machine has is able to run containerized apps with podman or docker.

1

Start your MCP-Server (Only necessary for On-Premise Setup)

a. Unzip the downloaded zip file.

b. Use the shipped docker-compose.yaml file or create one yourself.

c. Open the docker-compose.yaml file and check for any environment variables that you might need to replace

d. Start your container using the included start.bat or start.sh files or start the containerized compose yourself using: 

# Run on windows: 
start.bat
# Run on linux: 
start.sh
# Run with docker-compose: 
docker-compose up
# Run with podman-compose: 
podman-compose up

e. Verify the logs to see if your MCP-Server started successfully

2

Add the MCP-Server to your app

Use the following settings to connect with your MCP-client:

"command": "your-executable-path"

Transport Protocol: STDIO

3 snippets available

Claude Desktop (STDIO)

STDIO

Connect your STDIO MCP-Server to Claude Desktop

Requirements:

Claude Desktop is installed - Download it from: https://claude.ai/download

1

Prepare your MCP-Server

Depending on your operation system your zip file will include different executables. Make sure to unzip the whole folder.

You don´t have to start the executable file - The STDIO client will do that for you in the next steps!

Running on Windows

You are all set. Continue with step 2.

Running on Linux

Make the /lib/MCP-BuilderAI file executable.

chmod +x /path/to/your/lib/MCP-BuilderAI
Running on Mac

Make the /MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI file executable.

chmod +x /path/to/your/MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI

If you receive a message indicating that MCP-BuilderAI.app is invalid, it's likely because macOS has quarantined the executable.

Verify if your file has been quarantined.

xattr MCP-BuilderAI.app

Remove the quarantine flag from your executable

xattr -d com.apple.quarantine MCP-BuilderAI.app
2

Open Settings

Open Claude Desktop Settings

Claude Desktop Settings
3

Edit MCP-Server Config

Go to Developer -> Edit Config and Open the claude_desktop_config.json file.

Claude Desktop MCP-Server Settings
4

Add your MCP-Server to your Claude Instance

Add the following claude_desktop_config.json file.
If you have multiple MCP servers, list each one under the mcpServers array.

{
  "mcpServers": {
    "your-mcp-project": {
      "command": "/path/to/MCP-BuilderAI.exe",
      "env": {}
    }
  }
}

Depending on your operation system you might need to eccape your path using \\.

For example: C:\\Users\\....

5

Restart Claude Desktop

Close and restart Claude Desktop to load the new MCP server configuration.

To verify if the MCP-Server is connected ask Claude Desktop: "Which MCP Tools do you have?"

MCP-Inspector (STDIO)

STDIO

Debug your STDIO MCP-Server

Requirements:

npx is installed NPX

1

Prepare your MCP-Server

Depending on your operation system your zip file will include different executables. Make sure to unzip the whole folder.

You don´t have to start the executable file - The STDIO client will do that for you in the next steps!

Running on Windows

You are all set. Continue with step 2.

Running on Linux

Make the /lib/MCP-BuilderAI file executable.

chmod +x /path/to/your/lib/MCP-BuilderAI
Running on Mac

Make the /MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI file executable.

chmod +x /path/to/your/MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI

If you receive a message indicating that MCP-BuilderAI.app is invalid, it's likely because macOS has quarantined the executable.

Verify if your file has been quarantined.

xattr MCP-BuilderAI.app

Remove the quarantine flag from your executable

xattr -d com.apple.quarantine MCP-BuilderAI.app
2

Start & Open MCP Inspector

Open the terminal and run:

npx @modelcontextprotocol/inspector
Run MCP-Inspector

Get more infos at:MCP-Inspector

3

Connect with your STDIO Server

Enter the path to your exe file

MCP Inspector Dashboard

Others (STDIO)

STDIO

Connect your STDIO MCP-Server to your AI Application

Requirements:

Your AI application or SDK support MCP-client functionality. from: See which tools support MCP-Client

1

Prepare your MCP-Server

Depending on your operation system your zip file will include different executables. Make sure to unzip the whole folder.

You don´t have to start the executable file - The STDIO client will do that for you in the next steps!

Running on Windows

You are all set. Continue with step 2.

Running on Linux

Make the /lib/MCP-BuilderAI file executable.

chmod +x /path/to/your/lib/MCP-BuilderAI
Running on Mac

Make the /MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI file executable.

chmod +x /path/to/your/MCP-BuilderAI.app/Contents/MacOS/MCP-BuilderAI

If you receive a message indicating that MCP-BuilderAI.app is invalid, it's likely because macOS has quarantined the executable.

Verify if your file has been quarantined.

xattr MCP-BuilderAI.app

Remove the quarantine flag from your executable

xattr -d com.apple.quarantine MCP-BuilderAI.app
2

Add the MCP-Server to your app

Use the following settings to connect with your MCP-client:

"command": "your-executable-path"

Set the environment variables:

"env": ${}