Force Fabric MCP Server

A Model Context Protocol (MCP) server that provides live optimization analysis for Microsoft Fabric items. It connects to your Fabric tenant via Azure authentication and runs 100+ rules across Lakehouses, Warehouses, Eventhouses, and Semantic Models — detecting real issues with specific table and column names.

Features

🏠 Lakehouse Analysis (29 rules)

• REST API: SQL Endpoint status, Delta format check, medallion architecture naming
• SQL Endpoint: Data type analysis, nullable keys, empty tables, wide columns, naming conventions, audit columns, sensitive data
• OneLake Delta Log: VACUUM/OPTIMIZE history, auto-optimize settings, retention policies, file size analysis, write amplification, Z-Order, data skipping, partitioning

🏗️ Warehouse Analysis (39 rules)

• Schema: Primary keys, deprecated types, float precision, column/table naming, wide tables, foreign keys, circular FKs
• Data Quality: Nullable keys, empty tables, mixed date types, missing defaults, sensitive/PII columns
• Query Performance: Top slow queries, frequent queries, failed queries, volume trends, average duration
• Security: Data masking, RLS, privilege audit
• Database Config: AUTO_UPDATE_STATISTICS, result set caching, ANSI settings, snapshot isolation, page verify

📊 Eventhouse Analysis (13 rules per KQL database)

• Storage: Extent fragmentation, compression ratios, storage by table
• Policies: Caching, retention, merge, encoding, row order, partitioning, ingestion batching, streaming
• Health: Materialized views, data freshness, continuous exports, ingestion failures
• Queries: Performance summary (P95/avg/max), slow queries, failed commands

📐 Semantic Model Analysis (32 rules)

• DAX Expression Checks (via MDSCHEMA_MEASURES DMV): IFERROR, DIVIDE vs /, EVALUATEANDLOG, INTERSECT, duplicates, FILTER patterns, nested CALCULATE, SUMX, 1-(x/y), format strings
• Model Structure (via MDSCHEMA DMVs): Table count, date table, measure documentation, naming
• COLUMNSTATISTICS BPA (Import models): High-cardinality text, GUIDs, constants, boolean/date/number as text, string keys, wide tables, timestamps

Setup

Prerequisites

• Node.js 18+
• Azure CLI (az login) or another Azure authentication method
• Fabric capacity with items (Lakehouse, Warehouse, Eventhouse, or Semantic Model)

Install

git clone https://github.com/tmdaidevs/Force-Fabric-MCP-Server.git
cd Force-Fabric-MCP-Server
npm install
npm run build

Configure in VS Code

Add to your .vscode/mcp.json:

{
  "servers": {
    "fabric-optimization": {
      "type": "stdio",
      "command": "node",
      "args": ["dist/index.js"],
      "cwd": "${workspaceFolder}"
    }
  }
}

Or add to your global VS Code settings (settings.json):

{
  "mcp": {
    "servers": {
      "fabric-optimization": {
        "type": "stdio",
        "command": "node",
        "args": ["/absolute/path/to/Force-Fabric-MCP-Server/dist/index.js"]
      }
    }
  }
}

Usage

1. Authenticate

Use auth_login with method "azure_cli"

Available methods: azure_cli, interactive_browser, device_code, vscode, default, service_principal

2. List items

List all lakehouses in workspace <workspace-id>
List all warehouses in workspace <workspace-id>
List all eventhouses in workspace <workspace-id>
List all semantic models in workspace <workspace-id>

3. Run optimization scan

Run lakehouse optimization recommendations for <lakehouse-id> in workspace <workspace-id>
Run warehouse optimization recommendations for <warehouse-id> in workspace <workspace-id>
Run eventhouse optimization recommendations for <eventhouse-id> in workspace <workspace-id>
Run semantic model optimization recommendations for <model-id> in workspace <workspace-id>

Output Format

Every scan returns a unified rule results table:

15 rules — ✅ 9 passed | 🔴 1 failed | 🟡 5 warning

| Rule | Status | Finding | Recommendation |
|------|--------|---------|----------------|
| LH-007 Key Columns Are NOT NULL | 🔴 | 16 key column(s) allow NULL: ... | Add NOT NULL constraints... |
| LH-004 Table Maintenance | 🟡 | 4 Delta tables need OPTIMIZE... | Run lakehouse_run_table_maintenance... |

Only issues (FAIL/WARN) are shown in the table. Passed rules are counted in the summary.

Available Tools

Complete Rule Reference

🏠 Lakehouse Rules (29)

🏗️ Warehouse Rules (39)

📊 Eventhouse Rules (13 per KQL Database)

📐 Semantic Model Rules (32)

Architecture

src/
├── index.ts                 # MCP server entry point
├── auth/
│   └── fabricAuth.ts        # Azure authentication (CLI, browser, device code, SP)
├── clients/
│   ├── fabricClient.ts      # Fabric REST API + DAX executeQueries
│   ├── sqlClient.ts         # SQL endpoint via tedious (Lakehouse + Warehouse)
│   ├── kqlClient.ts         # KQL/Kusto REST API (Eventhouse)
│   ├── onelakeClient.ts     # OneLake ADLS Gen2 + Delta Log parser
│   └── xmlaClient.ts        # XMLA SOAP client (experimental)
└── tools/
    ├── ruleEngine.ts        # Shared RuleResult type + unified renderer
    ├── auth.ts              # Auth tools
    ├── workspace.ts         # Workspace tools
    ├── lakehouse.ts         # 29 rules (REST + SQL + Delta Log)
    ├── warehouse.ts         # 39 rules (SQL)
    ├── eventhouse.ts        # 13 rules per KQL DB (KQL)
    └── semanticModel.ts     # 32 rules (DAX + DMV + COLUMNSTATISTICS)

Authentication Methods

License

MIT