MCP Kusto

🚀 Azure Kusto MCP Server Configuration Guide

Azure Kusto Model Context Protocol (MCP) server is a powerful tool for interacting with Azure Kusto Query Language (KQL). This article will detail how to configure and use this server, including troubleshooting, best practices, and example queries.

🚀 Quick Start

Azure Kusto Model Context Protocol (MCP) server is a powerful tool for interacting with Azure Kusto Query Language (KQL). This guide will walk you through the configuration and usage of this server, including troubleshooting, best practices, and example queries.

📦 Installation

1. Create an MCP Configuration File

Create a file named mcp.json in the root directory of your project and add the following content:

{
    "servers": [
        {
            "name": "KustoMCP",
            "hostname": "<kusto-cluster-endpoint>",
            "port": 443,
            "database": "<database-name>",
            "table": "<table-name>",
            "enableTelemetry": true
        }
    ],
    "tracing": {
        "enabled": false,
        "samplingRate": 100
    },
    "tools": {
        "queryCompletion": {
            "enabled": true,
            "maxTokens": 500
        }
    }
}

2. Configure Visual Studio Code

In VS Code, navigate to the .vscode folder and create a new mcp.json file. Add the following content:

{
    "configurations": [
        {
            "name": "KustoMCP",
            "type": "ms-kusto",
            "request": "launch",
            "serverUrl": "<kusto-cluster-endpoint>",
            "databaseName": "<database-name>",
            "tableName": "<table-name>"
        }
    ]
}

💻 Usage Examples

1. Connect to the Cluster

Start a debugging session in VS Code:

# Run the following command in the terminal
az login
kustoctl --endpoint <kusto-cluster-endpoint> query "help;"

2. Execute Queries

Write and execute KQL queries directly in the VS Code editor. For example:

StormEvents
| where State == "FLORIDA"
| take 10

3. Analyze Data

Use the analyze command to analyze the results:

az kusto analyze --cluster <kusto-cluster-endpoint> --database <database-name> --query "<your-query-here>"

🔧 Technical Details

1. Configuration Issues

• Error: Unable to connect to the cluster
  Check the hostname and port in the mcp.json file.
• Error: MCP server not detected
  Ensure that the mcp.json file is located in the .vscode directory of your project.

2. Query Issues

• Error: Timeout
  Add filtering conditions or time limits to your queries.
• Error: Syntax error
  Use the optimize_query tool to check and fix query syntax.

📚 Documentation

Best Practices

1. Always connect first: Use the connect tool to establish a connection before executing queries.
2. Verify connection status: Use the connection_status tool to confirm that the connection is working properly.
3. Start small: Use the take or limit operators to limit the amount of data during testing.
4. Leverage analysis tools: The analyze_data tool can help you quickly obtain valuable information.
5. Optimize requests: Use the optimize_query tool to optimize complex queries.

KQL Query Examples

Basic Query

StormEvents
| where State == "FLORIDA"
| project TimeStamp, Severity, EventType

Time Series Analysis

HeartRateData
| where HeartRate > 100
| order by Timestamp asc

Data Aggregation

SalesData
| summarize TotalSales = sum(SalesAmount) by Region, Month

📄 Summary

With the guidance in this article, you can easily configure and use the Azure Kusto MCP server and master basic querying and analysis methods. When encountering issues, refer to the provided troubleshooting guide to resolve them.