MCP Kql Server

🚀 MCP KQL Server

A server based on the Microsoft Cloud Platform (MCP) for executing KQL queries against Azure Data Explorer through Azure CLI authentication.

🚀 Quick Start

📦 Installation

1. Make sure uv is installed:

pip install uv

2. Create a virtual environment and install dependencies:

uvx venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uvx pip install .

🏃‍♂️ Running the Server

Start the server using the following command:

uvx run start

The server will be available at http://0.0.0.0:5000.

💻 Usage Examples

🔐 Authentication

curl -X POST http://localhost:5000/auth

📝 Executing KQL Queries

curl -X POST http://localhost:5000/execute \
     -H "Content-Type: application/json" \
     -d '{"query": "cluster('\''mycluster'\'').database('\''mydb'\'').MyTable | take 10"}'

🛠️ Integration with Claude Desktop

Configure the server in claude_desktop_config.json to enable the kql_execute tool in Claude Desktop. Refer to mcp_config.yaml for detailed tool information.

🔧 Technical Details

📋 Requirements

• Python 3.9
• Azure CLI (az) for authentication
• Dependencies: Flask, waitress, azure-kusto-data, tenacity

⚠️ Troubleshooting

• Ensure that the Azure CLI is authenticated: az login
• Check the server logs for errors.
• Verify that port 5000 is open and accessible.