Kusto MCP Server

An MCP (Model Context Protocol) server that connects to your Azure Data Explorer (Kusto) database, exposing table schemas as resources and providing tools for data analysis.

Features

• Connection to Kusto Database: Securely connects to your Azure Data Explorer environment using service principal or managed identity authentication.
• Schema Resources: Exposes table schemas and sample data as resources for AI assistants to understand your data structure.
• KQL Query Tools: Provides tools for running read-only KQL queries against your database.
• Data Analysis Tools: Pre-built tools for common data analysis tasks like time series analysis, anomaly detection, etc.
• Analysis Prompts: Includes prompt templates for guiding AI assistants in performing common data analysis tasks.

Setup Instructions

Prerequisites

• Node.js 18.0 or higher
• Access to an Azure Data Explorer (Kusto) database
• Proper authentication credentials (service principal or managed identity)

Installation

1. Clone this repository or navigate to the project directory
2. Install dependencies:

npm install

3. Create a .env file based on the example:

cp .env.example .env

4. Configure the environment variables in the .env file:

# Kusto Connection Settings
KUSTO_CLUSTER_URI=https://your-cluster.kusto.windows.net
KUSTO_DATABASE=your-database

# Authentication settings 
# Option 1: Azure AD App Registration (Service Principal)
AZURE_CLIENT_ID=your-client-id
AZURE_CLIENT_SECRET=your-client-secret
AZURE_TENANT_ID=your-tenant-id

# Option 2: Managed Identity (if running in Azure)
# USE_MANAGED_IDENTITY=true

# MCP Server Config
MCP_SERVER_NAME=kusto-mcp-server
MCP_SERVER_VERSION=1.0.0

Build and Run

1. Build the TypeScript project:

npm run build

2. Start the MCP server:

npm start

Integrating with MCP Clients

Using with Claude Desktop

1. Configure Claude Desktop to use this MCP server by adding it to your claude_desktop_config.json file:

{
  "mcpServers": {
    "kusto-server": {
      "command": "node",
      "args": [
        "path/to/kusto-mcp-server/dist/index.js"
      ]
    }
  }
}

2. Start Claude Desktop and the server will automatically connect.

Using with Other MCP Clients

This MCP server uses the standard stdio transport, making it compatible with any MCP client that supports this transport.

Available Resources

• kusto://tables - Lists all available tables in the database
• kusto://schema/{tableName} - Shows schema information for a specific table
• kusto://sample/{tableName}/{sampleSize?} - Retrieves sample data from a specific table

Available Tools

• executeQuery - Executes a read-only KQL query against your database
• getTableInfo - Gets detailed schema and sample data for a specified table
• findTables - Finds tables that match a specified name pattern
• analyzeData - Performs various data analyses including summary statistics, time series analysis, top values analysis, outlier detection, and correlation analysis

Available Prompts

• Explore Dataset - Guide for exploring a new dataset
• Time Series Analysis - Analyze trends and patterns in time-based data
• Anomaly Detection - Identify unusual patterns or outliers in data
• Comparative Analysis - Compare data across different dimensions or time periods
• KQL Query Assistance - Get help building and optimizing KQL queries
• Data Quality Assessment - Evaluate and understand the quality of your data

Security Considerations

• This server only allows read-only operations to protect your data
• Contains basic security measures to prevent destructive operations
• Uses Azure AD authentication for secure database access
• Consider additional security measures depending on your specific requirements

License

ISC