Superset

You can connect clients to the Superset MCP Server over a stateless HTTP transport to manage dashboards, charts, datasets, and SQL execution programmatically. This MCP server lets you automate common Superset tasks through a consistent API, enabling chatbot or automation workflows to create dashboards, generate charts with auto-detected schema, and run SQL queries against your Superset instance.

How to use

You use an MCP client to communicate with the server via the /mcp endpoint exposed by the HTTP transport. Through this channel you can perform actions such as listing dashboards, creating dashboards, adding charts to dashboards, listing and creating charts (including auto-detection for schema), listing datasets and retrieving their schemas, and executing SQL queries directly against the connected Superset instance. The server handles authentication with Superset automatically and streams responses as needed.

Typical usage patterns include: selecting a target dashboard and adding charts, creating a new dashboard with a set of charts, auto-generating a chart from a dataset, and issuing SQL to fetch data for a visualization. When you retrieve data for a chart, you can rely on the server to provide the results in a structured form suitable for downstream rendering in your UI or chatbot.

How to install

Prerequisites you need before installing: Node.js v18 or newer, a running Apache Superset instance, and npm or Yarn.

# 1) Clone the MCP server repository
git clone <repository-url>
cd superset-mcp

# 2) Install dependencies
npm install

2. Configure environment variables. Copy the example environment file and customize it for your Superset instance and MCP server port.

cp .env.example .env

# Superset Configuration
SUPERSET_URL=http://localhost:8088
SUPERSET_USERNAME=admin
SUPERSET_PASSWORD=admin

# MCP Server Configuration
MCP_PORT=3000

3. Build and run the server. Build the TypeScript code, then start the server. For development with hot reload, you can use the dev script.

# Build TypeScript
npm run build

# Run the server
npm start

# Development with hot reload
npm run dev

4. Verify the server is up by checking the health endpoint. You should see a successful status and a list of available MCP tools.

curl http://localhost:3000/health

Additional configuration and notes

Open WebUI integration is supported by configuring a Tool Connection that points to the MCP server. You will add the MCP server using the following details in the admin tools area.

Configuration for Open WebUI (example)

Name: Superset MCP
Type: MCP (Streamable HTTP)
URL: http://YOUR_SERVER_IP:3000/mcp

Key endpoints exposed by the MCP server include the health check, the primary MCP endpoint for JSON-RPC requests, and a streaming endpoint for keep-alive messages.

Environment variables you may configure in your environment file (described above) include the Superset connection details and the MCP port. The server will automatically login to Superset and refresh tokens as needed.

Troubleshooting

Authentication issues usually indicate incorrect Superset credentials in your configuration. The server will attempt token refresh automatically when a token expires.

If you encounter CORS errors, ensure the server is allowed to accept requests from your client origin. Network firewalls should not block the MCP port.

If charts fail to display data, prefer using auto-detection when creating charts to improve metric and dimension inference.

API and endpoints overview

The server exposes health checks at /health and a primary MCP endpoint at /mcp for JSON-RPC requests. A streaming variant of /mcp is available to keep the connection alive.

Available tools

list_dashboards

List all dashboards available in Superset

get_dashboard

Get details of a dashboard including its charts

create_dashboard

Create a new dashboard

add_chart_to_dashboard

Add a single chart to a dashboard

add_charts_to_dashboard

Add multiple charts to a dashboard in one call (recommended)

list_charts

List all charts available in Superset

get_chart

Get details of a chart configuration

get_chart_data

Fetch data for a chart (chart data query)

create_chart

Create a new chart with manual configuration

create_chart_auto

Create a chart with auto-detection of schema and metrics

list_datasets

List all datasets or tables accessible to Superset

get_dataset_schema

Get the column definitions for a dataset

list_databases

List all database connections available to Superset

execute_sql

Execute a SQL query against a database connected to Superset