Spark SQL

Overview

Spark SQL is an MCP server that lets you query Spark SQL clusters via Thrift/HiveServer2. It works seamlessly with Spark, EMR, Hive, and Impala, exposing database operations as tools your AI assistant can invoke directly.

The server runs locally as a Python package. Install it with uvx spark-sql-mcp-server and configure it by setting the SPARK_HOST environment variable. You'll need an API key for authentication. Version v0.1.2 is available now under the MIT license, with full source code at aidancorrell/spark-sql-mcp-server on GitHub.

Spark SQL appears in the Database category on mcp.site and works with Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Zed, and other MCP clients. If you maintain this server, claim the listing to verify ownership and keep information up to date.

Installation

// claude_desktop_config.json
{
  "mcpServers": {
    "spark_sql": {
      "command": "uvx",
      "args": [
        "spark-sql-mcp-server"
      ],
      "env": {
        "SPARK_HOST": "<spark-host>"
      }
    }
  }
}

Requires SPARK_HOST