Superset MCP Server

A Model Context Protocol (MCP) server for managing Apache Superset datasets, metrics, and SQL queries.

> 📖 中文文档

🚀 Features

• Dataset Management: Full CRUD operations for Superset datasets
• Metrics Management: Create, update, and manage dataset metrics
• Calculated Columns: Create and manage calculated columns for datasets
• Chart Management: View and modify chart visualization parameters and filters
• Dashboard Operations: Access dashboard information, charts, and filters
• SQL Query Execution: Execute SQL queries directly through Superset
• Database Integration: List and manage database connections
• Resource Access: Browse datasets, databases, and metrics through MCP resources

📋 Prerequisites

• Node.js 18+
• Access to an Apache Superset instance
• Valid Superset credentials (username/password or access token)

🛠️ Installation

Using with Cursor or Claude Desktop

1. Add to MCP Configuration

Add the following configuration to your MCP settings file:

{
  "mcpServers": {
    "superset-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "superset-mcp"
      ],
      "env": {
        "SUPERSET_BASE_URL": "",
        "SUPERSET_USERNAME": "",
        "SUPERSET_PASSWORD": ""
      }
    }
  }
}

2. Environment Variables

Configure your Superset connection by updating the `env` section in the MCP configuration:

"env": {
  "SUPERSET_BASE_URL": "your-superset-url",
  "SUPERSET_USERNAME": "your_username",
  "SUPERSET_PASSWORD": "your_password",
}

Alternative: Using Access Token

"env": {
  "SUPERSET_BASE_URL": "your-superset-url",
  "SUPERSET_ACCESS_TOKEN": "your_access_token"
}

🔧 Available Tools

Dataset Operations

Metrics Operations

Calculated Columns Operations

Chart Operations

Dashboard Operations

SQL Operations

Database Operations

📚 Resources

Access read-only overviews through MCP resources:

• `superset://datasets` - Overview of all datasets
• `superset://databases` - List of database connections

Prompt examples

Use these natural prompts with your MCP-enabled assistant; it will pick the right tools and arguments.

• List datasets
  • "Show the first 10 datasets, most recently changed first. Only include id and table_name."

• Create a chart
  • "Create a simple table chart called 'Sample Table' using dataset 12."

• Update a chart
  • "Change chart 42 to a bar chart grouped by country and using SUM(value)."

• Dashboard query context
  • "On the 'sales-kpi' dashboard, show the full query context for chart 101."

• Run SQL
  • "On database 3, fetch the 10 most recently created users, returning only id and name."