microsoft_fabric_mcp

Overview

Microsoft Fabric MCP is a comprehensive Model Context Protocol server that connects AI assistants to Microsoft Fabric's data platform ecosystem. It provides 25 specialized tools for querying workspaces, examining table schemas, monitoring jobs, and analyzing dependencies without risking data modification. The server uses smart caching for performance and works seamlessly with Cursor, Claude, and other MCP-compatible AI tools. The implementation focuses on read-only operations using GET requests only, ensuring data safety while providing rich insights into Fabric resources. It includes detailed support for workspace management, data schema exploration, job monitoring, and operational intelligence features like lineage analysis and dependency mapping.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• list_workspaces

  List all accessible Fabric workspaces

  Try:List all my Fabric workspacesTry:Show me workspaces in Fabric
• get_workspace

  Get detailed workspace info including workspace identity status

  Try:Get details for 'DWH-PROD' workspaceTry:Show workspace info for ID 123
• list_items

  List all items in workspace with optional type filtering

  Try:List all items in 'Analytics' workspaceTry:Show me all lakehouses in workspace 456
• get_table_schema

  Get detailed schema for specific table

  Try:Get schema for 'sales' table in Bronze lakehouseTry:Show columns for 'customer' table
• list_job_instances

  List job instances with status/item filtering for monitoring

  Try:Show all running jobs in 'DWH-PROD'Try:List job instances for 'ETL_PROCESS'
• get_item_lineage

  Analyze data flow dependencies upstream/downstream

  Try:Show dependencies for 'sales_report' itemTry:Trace lineage of 'fact_orders'
• list_lakehouses

  List all lakehouses in specified workspace

  Try:List lakehouses in 'Analytics' workspaceTry:Show all my data lakehouses
• list_tables

  List all tables in lakehouse with format/type info

  Try:List all tables in 'Silver' lakehouseTry:Show tables in Bronze lakehouse
• get_job_instance

  Get detailed job info including errors and timing

  Try:Get details for job instance 789Try:Show errors for failed job 'ETL_Daily'
• list_compute_usage

  Monitor active jobs and estimate resource consumption

  Try:Show compute usage for last 24 hoursTry:List active resource-consuming jobs
• get_environment_details

  Get detailed environment config including Spark settings and libraries

  Try:Show details for 'dev' environmentTry:Get Spark settings for production env
• list_connections

  List all connections user has access to across entire tenant

  Try:Show all my Fabric connectionsTry:List data sources I can access

Comparable tools

Installation

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Windows
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

az login
az account set --subscription "Your-Subscription-ID"

"mcp_fabric": {
  "command": "uvx",
  "args": ["microsoft-fabric-mcp"]
}

"mcpServers": {
  "fabric": {
    "command": "uvx",
    "args": ["microsoft-fabric-mcp"]
  }
}

FAQ

Is this server safe to use with production data?

Yes, the server only uses GET requests for read-only operations, ensuring no data modification can occur.

What authentication method does this server use?

It uses Azure CLI authentication through DefaultAzureCredential, requiring proper Azure login before use.

Can I use this server with MCP tools other than Cursor?

Yes, it's designed to work with any MCP-compatible AI tools including Claude and other implementations.

Compare microsoft_fabric_mcp with