Fabric MCP Server is a Python-based MCP server for working with the Microsoft Fabric API. It lets an AI assistant manage Fabric workspaces, lakehouses, and warehouses, run SQL queries, and develop, test, and optimize PySpark notebooks, with LLM-assisted code generation and performance analysis built in.

Use it when you want an AI assistant to work directly against a Microsoft Fabric tenant: browsing and switching between workspaces, lakehouses, and warehouses, inspecting table schemas, running SQL, or building and tuning PySpark notebooks. It authenticates via Azure (az login), so it is only useful if you have Azure access to a Fabric tenant; it is not a general-purpose Spark or SQL tool for environments outside Microsoft Fabric.

• Manage workspaces, lakehouses, warehouses, and tables (list_workspaces, set_workspace, list_lakehouses, create_lakehouse, set_lakehouse, list_warehouses, create_warehouse, set_warehouse, list_tables, set_table)
• Retrieve Delta table schemas and metadata (get_lakehouse_table_schema, get_all_lakehouse_schemas)
• Execute SQL queries and load data (get_sql_endpoint, run_query, load_data_from_url)
• Perform operations on reports and semantic models
• Intelligent notebook creation with 6 specialized templates: basic, etl, analytics, ml, fabric_integration, and streaming
• Smart code generation for common PySpark operations
• Comprehensive validation with syntax checking and best practices
• Fabric-specific optimizations and compatibility checks
• Performance analysis with a 0-100 score and optimization recommendations
• Real-time monitoring and execution analytics

From source:

git clone https://github.com/your-repo/fabric-mcp.git
cd fabric-mcp
uv sync
pip install -r requirements.txt

Run it under the MCP Inspector for local development with uv run --with mcp mcp dev fabric_mcp.py, or as an HTTP server on port 8081 with uv run python .\fabric_mcp.py --port 8081. For VS Code, register it under STDIO or HTTP transport, for example:

{
    "mcp": {
        "servers": {
            "ms-fabric-mcp": {
                "type": "stdio",
                "command": "<FullPathToProjectFolder>\\.venv\\Scripts\\python.exe",
                "args": ["<FullPathToProjectFolder>\\fabric_mcp.py"]
            }
        }
    }
}

It requires Azure authentication via az login --scope https://api.fabric.microsoft.com/.default. Both STDIO and HTTP communication modes are supported.

Typical requests it is built to handle include listing all workspaces, creating a PySpark notebook that reads sales data, cleans it, and optimizes performance, or diagnosing a slow PySpark notebook and suggesting fixes.

Data engineers and analysts working in Microsoft Fabric who want an AI assistant to manage Fabric resources and write, validate, and optimize PySpark notebooks directly, instead of switching between the Fabric UI and a notebook editor by hand.