pbi-desktop -public

This MCP server enables AI assistants to read and modify Power BI Desktop files locally. It exposes capabilities for querying the model, creating measures, managing relationships, and optimizing performance through natural language prompts. Once running, you can connect an MCP-compatible client (such as Claude Desktop, Copilot, or ChatGPT) and issue natural-language instructions to perform tasks within Power BI Desktop, with the server translating those instructions into concrete Power BI actions. The server operates locally on your machine, ensuring data never leaves your device unless you choose to share it with a provider through the client. Typical workflows include asking the model to examine a model, add a measure, or refactor relationships, and then applying those changes through Power BI Desktop in a controlled, test-before-deploy manner.

Prerequisites:

• A supported host OS (Windows 10/11 or macOS)
• Node.js installed on your system (recommended LTS version)
• Access to the MCP Engine locally (download the release or build from source as provided by the project)

Installation steps:

1. Install Node.js if you don’t have it:

   • macOS/Linux: curl -fsSL https://nodejs.org/dist/v18.17.0/node-v18.17.0.pkg | sudo installer -pkg -target /
   • Windows: download and run the Node.js installer from https://nodejs.org/

2. Download the latest MCP Engine release for pbi-desktop-public (from the official repository or release page).

3. Prepare the server configuration file (example using the included server script path):

   • Ensure you have a server.js (or equivalent) that starts the MCP server.
   • If using a package, install dependencies as needed (e.g., npm install).

4. Start the MCP server:

   • npm-based launch (if applicable): npm install && npm run start
   • Direct Node.js launch (alternative): node path/to/server.js

5. Verify the server is listening on the configured port (default 8080) and accessible from your MCP client.

6. Connect your MCP client and authorize any required credentials per the client’s documentation.

Notes and tips:

• The server runs locally to protect data; ensure your firewall allows the chosen port if you plan to access it from a client on the same machine.
• Environment variables can control logging and port configuration. Example: PBI_DESKTOP_PORT=8080, MCP_LOG_LEVEL=info/debug.
• If you encounter connection issues, verify that the server.js (or equivalent) is using the same port as configured in the client.
• This MCP server supports popular AI assistants (Claude, Copilot, ChatGPT) via MCP clients; refer to the client’s integration guide for connection details.
• Since data can interact with AI providers depending on prompts, review privacy settings and keep sensitive data local when possible.