Edit

Get started with Fabric MCP Server (local)

Install the local Fabric MCP Server and use AI agents to work with Fabric APIs, OneLake data, and item definitions from your development machine.

Note

Fabric MCP Server is an open-source project on GitHub. This quickstart covers the essentials — the GitHub repository has detailed setup instructions and troubleshooting.

Prerequisites

Before you install Fabric MCP Server, make sure you have:

  • An MCP-compatible AI agent — VS Code with GitHub Copilot (recommended) or any MCP host that supports stdio transport

Additional prerequisites depend on your installation method:

Method Prerequisite
VS Code extension None
npm/npx Node.js 20 LTS or later (download)
Source build .NET 9 SDK or later (download)

Install

Works with both the Stable and Insiders builds of VS Code.

  1. Install the GitHub Copilot Chat extension.
  2. Install the Fabric MCP Server extension.

You don't need any further configuration. The extension registers the MCP server automatically.

Option 2: npm/npx (any MCP client)

Requires Node.js 20 LTS or later. Verify with node --version.

Add the following to your client's mcp.json file:

{
    "mcpServers": {
        "fabric-mcp-server": {
            "command": "npx",
            "args": [
                "-y",
                "@microsoft/fabric-mcp@latest",
                "server",
                "start",
                "--mode",
                "all"
            ]
        }
    }
}

Option 3: Build from source (.NET)

Requires .NET 9 SDK or later. Verify with dotnet --version.

git clone https://github.com/microsoft/mcp.git
cd mcp
dotnet build servers/Fabric.Mcp.Server/src/Fabric.Mcp.Server.csproj --configuration Release

Then add the following to your client's mcp.json file:

{
    "mcpServers": {
        "Fabric MCP Server": {
            "command": "/path/to/repo/servers/Fabric.Mcp.Server/src/bin/Release/fabmcp",
            "args": [
                "server",
                "start"
            ],
            "type": "stdio"
        }
    }
}

Note

On Windows, use backslashes and add the .exe extension: C:\path\to\repo\servers\Fabric.Mcp.Server\src\bin\Release\fabmcp.exe

Note

When you configure Visual Studio or Visual Studio Code manually, use servers instead of mcpServers as the root JSON key.

Client-specific configuration

If you didn't use the VS Code extension, place your mcp.json configuration in the location required by your MCP host:

Client File location Documentation
VS Code .vscode/mcp.json (workspace) or settings.json (user) VS Code MCP docs
Other hosts Refer to your host's documentation for the MCP server configuration file location —

Verify your installation

  1. Open GitHub Copilot in VS Code and switch to Agent mode.

  2. Select refresh on the tools list. You should see the Fabric MCP Server tools listed.

  3. Try a prompt that uses Fabric context:

    What Fabric workload types are available?

If the agent returns a list of Fabric workload types, your server is working correctly.

Example prompts

After installation, try these prompts to explore what the server can do:

Scenario Example prompt
Browse API specs "Show me the OpenAPI operations for 'notebook' and give a sample creation body"
Explore item schemas "Show me the JSON schema for a Data Pipeline item definition"
Get best practices "Show me best practices for handling API throttling in Fabric"
List workspaces "List my Fabric workspaces"
Work with files "Download the config.json file from my Sales lakehouse"
Create items "Create a new Lakehouse called CustomerData in my dev workspace"

For a complete list of tools and multi-step workflow examples, see the tools reference.

Next steps

Related content

  • Last updated on