SSAS MCP Server: Connect AI Agents to Your Analysis Services Cubes

Your SSAS cubes hold some of the most valuable data in your organization. But querying them still requires DAX or MDX expertise. What if Claude, GitHub Copilot, or any AI assistant could just talk to your cubes directly?

That's exactly what the SSAS MCP Server does. It's an open source, Python-based MCP server that connects AI agents to SQL Server Analysis Services. You install it with pip, point it at your SSAS instance, and start asking questions in plain English.

We built it at Nexus AI because we kept running into the same problem with our clients: the data is there, the cubes are well-modeled, but only a handful of people in the organization can actually query them. That's a bottleneck that shouldn't exist anymore.

Why SSAS Still Matters (and Why It Needs AI)

SSAS has been the backbone of enterprise analytics for over two decades. Thousands of organizations rely on it for financial reporting, sales analysis, and operational dashboards. The data models are mature, well-governed, and trusted.

But here's the thing: DAX and MDX are not easy. They're powerful query languages, but they have a steep learning curve. The result is that most business users depend on pre-built reports and dashboards. When they need a quick answer that falls outside an existing report, they file a request and wait.

Meanwhile, AI assistants have gotten remarkably good at understanding data and generating queries. The missing piece was a standardized way to connect them to your data sources. That's where MCP comes in.

What is MCP?

The Model Context Protocol (MCP) is an open standard developed by Anthropic that defines how AI assistants interact with external tools and data. Think of it as a universal adapter between AI models and the systems they need to work with.

The protocol has grown rapidly since its release in November 2024. There are now over 10,000 MCP servers available, covering everything from GitHub and Slack to databases and file systems. Major players like Microsoft, Google, and Anthropic all support MCP in their AI tools.

For BI teams, MCP means one thing: your AI assistant can now query your data directly, without custom integrations or middleware. It discovers the available tools, understands the schema, generates the right queries, and returns structured results.

What the SSAS MCP Server Does

The server exposes six tools that give AI agents full read access to your SSAS instance:

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

• "What were our top 10 products by revenue last quarter?"

• "Show me all the measures in the Sales model and their DAX expressions"

• "Compare gross margin across regions for 2024 vs 2025"

• "Which dimensions are available in the Finance cube?"

The AI agent handles the rest. It discovers the model structure, generates the appropriate DAX or MDX query, executes it, and presents the results in a readable format.

How It Compares to Alternatives

There are a few other ways to connect AI to your analytics data. Here's how the SSAS MCP Server stacks up:

A few things stand out.

CData offers a commercial SSAS MCP server, but it requires a paid license and runs on Java. It translates your queries to SQL rather than using native DAX/MDX, which can limit the analytical depth. If you're already paying for CData drivers across your stack, it might make sense. Otherwise, it's a lot of overhead for what should be a simple connection.

Microsoft's Power BI MCP Server is great if you're fully in the Fabric/Power BI ecosystem. But it only works with Power BI semantic models, not standalone SSAS instances. If your cubes run on SSAS (on-premises or Azure Analysis Services), this one won't help you.

Microsoft's SQL MCP Server is built for relational databases. It doesn't support OLAP, DAX, or MDX. Different tool for a different job.

The SSAS MCP Server fills the gap: a lightweight, open source, Python-native solution that talks directly to your SSAS instance using native DAX and MDX. No middleware, no paid licenses, no Java dependencies.

Getting Started

You need three things: Python 3.10+, the ADOMD.NET client library (ships with SSMS or Power BI Desktop), and a Windows machine (ADOMD.NET requires .NET Framework).

Install

Configure

set SSAS_SERVER=YOUR_SERVER\INSTANCE
set SSAS_DATABASE

set SSAS_SERVER=YOUR_SERVER\INSTANCE
set SSAS_DATABASE

set SSAS_SERVER=YOUR_SERVER\INSTANCE
set SSAS_DATABASE

Connect to Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

{
  "mcpServers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

{
  "mcpServers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

Connect to VS Code / GitHub Copilot

Add this to your .vscode/mcp.json:

{
  "servers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

{
  "servers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

{
  "servers": {
    "ssas": {
      "command": "python",
      "args": ["-m", "ssas_mcp_server"],
      "env": {
        "SSAS_SERVER": "YOUR_SERVER\\INSTANCE",
        "SSAS_DATABASE": "Your Cube Name"
      }
    }
  }
}

Connect to Claude Code

claude mcp add ssas -- python -m

claude mcp add ssas -- python -m

claude mcp add ssas -- python -m

That's it. Open your AI client, and start asking questions about your data.

Real-World Use Cases

We've seen this tool make an immediate difference in a few scenarios:

Ad-hoc reporting during meetings. Instead of saying "I'll get back to you with those numbers," the analyst pulls up Claude and asks the question right there. The answer comes back in seconds, with the actual DAX query visible for verification.

Data model exploration. New team members or consultants can ask "describe the data model" and get a structured overview of all tables, measures, and relationships. No more digging through documentation that may or may not be up to date.

Rapid prototyping of reports. Before building a full Power BI report, you can test whether the data supports the analysis you have in mind. Ask a few questions, validate the measures, and then build with confidence.

Self-service for non-technical stakeholders. Finance directors and product managers don't need to learn DAX. They describe what they need in plain English, and the AI generates the query. The technical team can review the generated DAX if needed.

Security Considerations

A fair question: should you give an AI agent access to your SSAS data?

The SSAS MCP Server is read-only by design. It can execute DAX and MDX queries, but it cannot modify data, alter the model, or change any configuration.

Authentication follows your existing SSAS instance security. The server connects using the credentials of the Windows user running the process. If that user only has read access to specific cubes, the AI agent inherits those same restrictions.

The server runs locally via stdio transport. There's no network endpoint exposed, no API to secure, no cloud service involved. Your data stays where it is.

That said, use common sense. Don't run it with admin credentials. Restrict access to the cubes and measures that are appropriate for the use case. Treat it like you would any other reporting tool.

Built and Maintained by Nexus AI

We built this at Nexus AI as part of our work on AI-powered automation for business. We help organizations connect their existing data infrastructure to modern AI tools, without ripping out what already works.

The SSAS MCP Server is open source and free to use. You can find the full source code, contribute, or report issues on GitHub.

If you're looking to integrate AI agents into your BI stack, or you need help setting up MCP servers for your specific data sources, get in touch. We work with organizations across the Netherlands and beyond to make their data more accessible through AI.

What's Next

We're actively developing the SSAS MCP Server. Some things on our radar:

• Broader platform support for Linux/macOS environments

• Enhanced model metadata for richer AI context

• Query validation to catch common DAX/MDX mistakes before execution

• Integration guides for more MCP clients as the ecosystem grows

Want to stay updated or contribute? Star the GitHub repository and watch for new releases.

If your organization runs SSAS and you want to explore what AI can do with your data, we'd love to hear from you. Reach out to Nexus AI for a no-obligation conversation about your setup.