Frequently Asked Questions
Where do you get MCP Server from and what are the installation requirements?
The MCP Server can be launched individually or as part of the RAG-in-a-box system. It is distributed as pre-compiled binaries that can run on various operating systems, including:
Windows
RHEL (Red Hat Enterprise Linux)
Ubuntu
Is MCP Server a command-line tool, or does it have a GUI?
The MCP Server is a network service that runs as an HTTP server; it does not have a graphical user interface (GUI) or a direct command-line interface (CLI) for tools. It's designed to be a backend service that is:
Accessed programmatically via the Model Context Protocol.
Integrated into AI assistants and clients like Claude Desktop, Cursor, or Windsurf.
You interact with the server by configuring a client application to communicate with it. For example, here is how you might configure a client like Windsurf:
{
"mcpServers": {
"rag-mcp": {
"serverUrl": "http://localhost:8002/mcp",
"headers": {
"Authorization": "Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.xyz.abc"
}
}
}
}How do you configure the MCP Server and connect it to MariaDB?
The MCP Server does not include its own database. It acts as a client and requires a connection to an external, pre-existing MariaDB server.
The system components are connected as follows:
MCP Server (Port 8002) ---------> MariaDB Server (Port 3306)
(connects via MySQL protocol)Configuration is managed through environment files where you specify the connection details for your MariaDB instance.
How are tools like list_databases executed?
Tools are not typed into a command line. Instead, they are executed programmatically by a Large Language Model (LLM) in response to a user's query in natural language.
The process works like this:
A user asks a question in an integrated client (e.g., "Can you show me what databases are available?").
The LLM interprets the request and determines that the
list_databasestool is needed.The LLM calls the
list_databasestool by sending a JSON-RPC request to the MCP Server.The MCP Server executes the tool against the connected MariaDB database.
The results are sent back to the LLM, which formats them into a natural language response for the user.
What are the JSON snippets in the documentation for?
The JSON snippets shown in the documentation are examples of the "behind-the-scenes" communication between a client, the LLM, and the MCP Server. They are not meant to be copied and pasted into a CLI but serve to illustrate how the protocol functions.
This page is: Copyright © 2025 MariaDB. All rights reserved.
Last updated
Was this helpful?

