kusto

This MCP server wraps the Kusto (Azure Data Explorer) integration so your AI assistant can connect to your Azure Data Explorer clusters and run queries using natural language. With the kusto-mcp, your AI partner can initialize connections, browse database tables and schemas, and execute KQL queries without you needing to write KQL yourself. It exposes commands that let you connect to a cluster, list tables, inspect schemas, and run queries, returning results formatted for AI consumption. The included tools are designed to be discoverable by compatible MCP clients (Claude Code, Cline, Cursor, Claude Desktop, VS Code MCP, and other MCP-enabled tools).

Prerequisites:

• Node.js and npm installed on your machine (or your environment where MCPs run).
• Azure CLI installed and a valid Azure account to access your Data Explorer clusters.

Installation steps:

1. Prepare your environment:

   • Ensure you are logged in to Azure after installing Azure CLI: az login

2. Install or run the MCP server via npx (preferred for quick setup):

   • This fetches the latest kusto-mcp package and runs it directly: npx -y kusto-mcp@latest

3. Optional: wire into a client configuration (example for GitHub-style MCP config):

   • You can place a configuration snippet like the following in your cline_mcp_settings.json or VS Code MCP settings to point to the server: { "mcpServers": { "github.com/johnib/kusto-mcp": { "command": "npx", "args": ["-y", "kusto-mcp@latest"], "env": {}, "disabled": false, "autoApprove": [ "initialize-connection", "show-tables", "show-table", "execute-query" ] } } }

4. Validate the connection:

   • Use your MCP client to invoke a connection to your Azure Data Explorer cluster and list available tables, then run a sample query.

Tips and common issues:

• Ensure you are authenticated to Azure (az login) and have access to the target Azure Data Explorer cluster.
• If the AI tool cannot see the tools, restart the AI assistant after adding the configuration and verify that the MCP JSON is valid.
• When configuring in various clients (Cline, Cursor, Claude Desktop, etc.), use the provided sample settings to ensure the server is reachable and the correct command is used (npx with kusto-mcp@latest).
• The MCP server handles initialization, table discovery, and KQL query execution, and formats results for AI-friendly responses to avoid overwhelming the context window.
• If you experience connectivity issues, double-check cluster URL formats and network access to Azure Data Explorer.

Environment variables (optional):

• You can add environment-specific overrides in the env object within your client config if needed (e.g., for proxy settings or cluster-specific flags).

For more details, consult the project's configuration and troubleshooting documentation.