Microsoft Fabric

You can run, configure, and interact with the Microsoft Fabric MCP Server to manage Fabric workspaces, lakehouses, tables, and pipelines, while leveraging PySpark notebook development, validation, and optimization features powered by AI-assisted guidance. This server exposes a practical set of operations for Fabric resources, along with templates, code generation, and performance insights to speed up your data engineering workflows.

How to use

You use an MCP client to connect to the MCP Server either through stdio (local server) or HTTP (remote server). Start a local MCP server instance and then connect from your development environment (IDE, CLI, or your custom client). The server provides functions to manage workspaces, lakehouses, warehouses, tables, notebooks, and reports, and it can generate PySpark code, validate syntax, and analyze performance.

How to install

Prerequisites you need before installing the MCP server:

• Python 3.12+

• Azure credentials for authentication

• uv (from astral) for server runner

• Azure CLI for interacting with Fabric APIs

• Optional: Node.js if you plan to use the MCP inspector in your workflow

Configuration and usage notes

The MCP server is designed to operate with a PySpark-oriented Fabric workflow. It integrates a set of tools to interact with Fabric resources, generate code, validate syntax, and analyze performance. When running locally, you will start the server in stdio mode or via HTTP depending on your setup. Administrator and developer credentials should be secured and managed per your organization’s security policies.

Troubleshooting

If you encounter issues, check authentication state with your Azure credentials, ensure you have the correct workspace context, and verify that your server instance is running on the expected port. Use code validation tools to catch syntax or best-practice issues, and review performance analysis results to identify bottlenecks or optimization opportunities.

Notes on usage patterns

Typical usage includes creating notebooks from templates, generating PySpark code for common operations, validating Fabric-compatible code, and deploying updates to Fabric resources. You can combine natural language prompts with the AI-assisted workflow to create Fabric-optimized notebooks, run SQL against lakehouses and warehouses, and load data from URLs into Delta tables.

Examples of common workflows

Create a PySpark notebook from a template, connect to a lakehouse, read data, perform cleaning steps, and apply performance optimizations. Validate the notebook code, analyze performance, and iterate with AI-driven suggestions until the workflow meets your requirements.

Security considerations

Always protect Azure credentials and access keys. Use least-privilege permissions for Fabric resources, rotate tokens, and follow your organization’s security guidelines when exposing MCP endpoints to developers.

Available tools

MCP Tools

PySpark tools for code generation, validation, and Fabric interactions (PySpark Tools, PySpark Helpers, Template Manager, Code Validators, Code Generators) making it easier to build, validate, and optimize PySpark workflows.

Code Validators

Validate PySpark and Fabric-compatible code for syntax, best practices, and Fabric compatibility before deployment.

Code Generators

Automatically generate PySpark code and Fabric-specific snippets tailored to common operations and templates.

Template Manager

Create and manage notebook templates for basic, ETL, analytics, fabric integration, and streaming workflows.

Notebook Utilities

Utilities to create, inspect, and manage Fabric-optimized notebooks with guidance on performance patterns.