SSAS MCP Server

A Model Context Protocol (MCP) server that connects AI agents to SQL Server Analysis Services (SSAS). It lets tools like GitHub Copilot, Claude (via Codex or Claude Code), and other MCP clients run DAX and MDX queries, explore models, and list measures, all through natural language.

Windows only
SSAS uses the ADOMD.NET client library, which requires .NET Framework on Windows.

Features

Prerequisites

1. Python 3.10+

2. ADOMD.NET client library
   The DLL Microsoft.AnalysisServices.AdomdClient.dll must be present on the machine. It ships with any of these:

   • SQL Server Management Studio (SSMS)
   • Power BI Desktop
   • The standalone AMO/ADOMD client libraries

   The server auto-detects common install locations. If your DLL is elsewhere, set the ADOMD_DLL_PATH environment variable to the folder containing it.

Installation

pip install ssas-mcp-server

After installation, the server can be started with:

ssas-mcp-server

Or install from source:

git clone https://github.com/NexusAI-Solutions/ssas-mcp-server.git
cd ssas-mcp-server
pip install .

Configuration

The server is configured through environment variables:

Usage

Run directly

set SSAS_SERVER=SERVER\INSTANCE
set SSAS_DATABASE=My Cube
ssas-mcp-server

The server starts in stdio mode and waits for an MCP client to connect.

You can also run it as a Python module:

python -m ssas_mcp_server

VS Code / Codex

Add this to your .vscode/mcp.json:

{
  "servers": {
    "ssas": {
      "command": "ssas-mcp-server",
      "env": {
        "SSAS_SERVER": "SERVER\\INSTANCE",
        "SSAS_DATABASE": "My Cube"
      }
    }
  }
}

Claude Code

claude mcp add ssas -- ssas-mcp-server

Then set the environment variables in your shell before starting Claude Code, or pass them inline.

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "ssas": {
      "command": "ssas-mcp-server",
      "env": {
        "SSAS_SERVER": "SERVER\\INSTANCE",
        "SSAS_DATABASE": "My Cube"
      }
    }
  }
}

Example queries

Once connected, you can ask your AI assistant things like:

• "List all tables in the SSAS model"
• "Show me the measures and their DAX expressions"
• "Run this DAX query: EVALUATE TOPN(10, 'Sales', [Revenue], DESC)"
• "Describe the data model"
• "What columns does the Customers table have?"

Troubleshooting

Could not load file or assembly 'Microsoft.AnalysisServices.AdomdClient'

The ADOMD.NET DLL was not found. Solutions:

1. Install SSMS, Power BI Desktop, or the standalone AMO client libraries.
2. Set ADOMD_DLL_PATH to the folder containing Microsoft.AnalysisServices.AdomdClient.dll.

To find the DLL on your system:

where /r "C:\Program Files" Microsoft.AnalysisServices.AdomdClient.dll

Server is not found or not accessible

• Verify the server name matches exactly what works in SSMS or Power BI.
• For named instances, use the SERVER\INSTANCE format.
• Make sure the SQL Server Browser service is running if you are connecting by instance name.

License

MIT

Publishing to PyPI

Install the packaging tools:

python -m pip install --upgrade build twine

Build and validate the distribution files:

python -m build
python -m twine check dist/*

Upload to TestPyPI first:

python -m twine upload --repository testpypi dist/*

If the TestPyPI package installs and runs correctly, upload to PyPI:

python -m twine upload dist/*

Automatic publishing from GitHub

This repository includes a GitHub Actions workflow that builds and publishes the package to PyPI when a GitHub Release is published.

To enable it, create a PyPI Trusted Publisher for this project:

• PyPI project name: ssas-mcp-server
• Owner: NexusAI-Solutions
• Repository name: ssas-mcp-server
• Workflow name: publish.yml
• Environment name: pypi

Then publish a GitHub Release after bumping the version in pyproject.toml. PyPI does not allow replacing an existing version, so every release must use a new version number.