MCP Server Setup for Power BI 2026

20 Comments / Blog / By

You are installing the Power BI Modeling MCP Server — a local server that allows VS Code + GitHub Copilot to connect to a running Power BI Desktop model so your AI assistant can read and interact with the semantic model. (GitHub)

🛠️ Prerequisites (Before You Start)

Make sure you have:

✔️ Windows OS (MCP for Desktop works only on Windows)
✔️ Power BI Desktop installed
✔️ Visual Studio Code
✔️ GitHub Copilot + GitHub Copilot Chat extensions in VS Code

📦 Step 1 — Install the MCP Server

✅ Option A — Use VS Code Integration (Recommended)

  1. Open Visual Studio Code
  2. Go to Extensions
  3. Search and install:
    • Power BI Modeling MCP extension
    • GitHub Copilot + Copilot Chat
  4. After installation open the GitHub Copilot Chat panel and confirm that powerbi-modeling-mcp appears as a registered MCP server option. (GitHub)

This method manages the MCP server for you inside VS Code.

🧠 Step 2 — Make Sure Power BI Desktop Is Running

  1. Open Power BI Desktop
  2. Load your .pbix file
  3. Keep the file open — the MCP server will auto-discover the model when connecting later.

⚙️ Step 3 — Enable GitHub Copilot Agent Mode

  1. In VS Code, open GitHub Copilot Chat
  2. Click the Tools / Agent Mode button to activate agent-style interactions
  3. This lets Copilot use the MCP tools to communicate with your Power BI model.

Copilot + its MCP toolset must be running together so that the MCP server appears active when you start querying. (GitHub)

🧪 Step 4 — Connect to Your Power BI Model

With VS Code + Copilot Chat open:

  1. Ask a simple prompt like:List all tables in my open Power BI model
  2. Copilot will use the MCP server to detect the open desktop instance and return the list.

Once connected, you can use natural language to:

✔ list tables, columns, measures
✔ run or validate DAX queries
✔ get column details, relationships, model stats
✔ ask Copilot to generate or refactor DAX code based on your prompts

The MCP server handles all the connection and tooling protocols under the hood. (GitHub)

🧹 Optional Manual MCP Server Install

If you don’t want to use the VS Code extension:

  1. Download the latest .vsix from the marketplace
  2. Rename it to .zip and extract
  3. Run the included powerbi-modeling-mcp.exe
  4. Register it manually in VS Code’s MCP config (.vscode/mcp.json) with:
{
  "servers": {
    "powerbi-modeling-mcp": {
      "type": "stdio",
      "command": "C:\\MCPServers\\PowerBIModelingMCP\\extension\\server\\powerbi-modeling-mcp.exe",
      "args": ["--start"],
      "env": {}
    }
  }
}

This is only needed if you can’t use the VS Code extension. (GitHub)

🎯 Quick Troubleshooting Tips

✔ Make sure Power BI Desktop is open with your report loaded
✔ Restart VS Code after installing the MCP extension
✔ Enable Agent Mode in GitHub Copilot Chat
✔ Confirm Copilot + MCP server appears in the chat tools panel

If Copilot doesn’t show an MCP connection, reload the window. (GitHub)

🔖 Summary

✅ Install VS Code + GitHub Copilot extensions
✅ Install the Power BI Modeling MCP extension
✅ Open your Power BI Desktop file
✅ Enable Agent Mode in Copilot Chat
✅ Use natural language prompts to explore or modify your model

This setup lets your AI-assisted development workflows connect directly to your local Power BI Desktop semantic model. (GitHub)

#PowerBI #MCPServer #PowerBIDesktop #VSCode #GitHubCopilot #DataModeling #SemanticModels #DAX #AIforBI #DataAnalytics #Datatipss

Related Posts

Subscribe
Subscribe