MCP Apache Superset
Connects LLM clients to Apache Superset for querying charts, dashboards, datasets, databases, and executing SQL.
README
MCP Apache Superset
A minimal Model Context Protocol (MCP) server that connects LLM clients (Claude Code, Claude Desktop, Cursor, etc.) to a local or remote Apache Superset instance.
Table of Contents
- Features
- Prerequisites
- Installation
- Configuration
- Authentication
- Usage
- Available Tools
- Running Superset Locally
- Architecture
Features
- 12 tools covering charts, dashboards, datasets, databases, SQL execution, and query history
- Automatic authentication with token refresh on 401
- CSRF token handling for write operations
- Supports database auth, OAuth2/OIDC SSO, LDAP, and pre-authenticated access tokens
- Zero external dependencies beyond the MCP SDK and Zod
Prerequisites
- Node.js >= 18
- Apache Superset running and accessible (local or remote)
Installation
git clone <this-repo>
cd mcp-apache
npm install
npm run build
Configuration
All configuration is done via environment variables:
| Variable | Default | Description |
|---|---|---|
SUPERSET_URL |
http://localhost:8088 |
Superset base URL |
SUPERSET_USERNAME |
admin |
Username for database/LDAP auth |
SUPERSET_PASSWORD |
admin |
Password for database/LDAP auth |
SUPERSET_ACCESS_TOKEN |
(none) | Pre-authenticated JWT access token (skips login) |
SUPERSET_AUTH_PROVIDER |
db |
Auth provider: db, ldap, oauth, or token |
Copy the example file to get started:
cp .env.example .env
Authentication
The MCP server supports multiple authentication methods to match your Superset deployment.
Database Auth (Default)
The simplest method. Superset stores users in its own metadata database.
SUPERSET_AUTH_PROVIDER=db
SUPERSET_USERNAME=admin
SUPERSET_PASSWORD=admin
The server calls POST /api/v1/security/login with provider "db" and receives a JWT access token.
SSO / OAuth2 / OIDC
When Superset is configured with OAuth2 or OpenID Connect (Azure AD, Okta, Keycloak, Google, etc.), you cannot log in via the REST API directly because the flow requires a browser redirect. There are two approaches:
Option A: Use a pre-authenticated access token
This is the recommended approach for MCP servers. You obtain a token outside the MCP flow and pass it directly.
Step 1: Configure Superset for OAuth2/OIDC in superset_config.py:
from flask_appbuilder.security.manager import AUTH_OAUTH
AUTH_TYPE = AUTH_OAUTH
OAUTH_PROVIDERS = [
{
"name": "keycloak", # or "azure", "okta", "google", etc.
"icon": "fa-key",
"token_key": "access_token",
"remote_app": {
"client_id": "superset-client",
"client_secret": "YOUR_CLIENT_SECRET",
"server_metadata_url": "https://keycloak.example.com/realms/myrealm/.well-known/openid-configuration",
"api_base_url": "https://keycloak.example.com/realms/myrealm/protocol/openid-connect",
"access_token_url": "https://keycloak.example.com/realms/myrealm/protocol/openid-connect/token",
"authorize_url": "https://keycloak.example.com/realms/myrealm/protocol/openid-connect/auth",
"client_kwargs": {
"scope": "openid email profile"
},
},
}
]
# Map OAuth roles to Superset roles
AUTH_ROLES_MAPPING = {
"superset_admin": ["Admin"],
"superset_alpha": ["Alpha"],
"superset_gamma": ["Gamma"],
}
AUTH_ROLES_SYNC_AT_LOGIN = True
Step 2: Obtain a token via OAuth2 Client Credentials or Resource Owner Password Grant (for service accounts):
# Client Credentials Grant (service-to-service, no user context)
curl -X POST https://keycloak.example.com/realms/myrealm/protocol/openid-connect/token \
-d "grant_type=client_credentials" \
-d "client_id=superset-client" \
-d "client_secret=YOUR_CLIENT_SECRET"
# Resource Owner Password Grant (if enabled by your IdP)
curl -X POST https://keycloak.example.com/realms/myrealm/protocol/openid-connect/token \
-d "grant_type=password" \
-d "client_id=superset-client" \
-d "client_secret=YOUR_CLIENT_SECRET" \
-d "username=svc-mcp" \
-d "password=SVC_PASSWORD" \
-d "scope=openid"
Step 3: Pass the token to the MCP server:
SUPERSET_AUTH_PROVIDER=token
SUPERSET_ACCESS_TOKEN=eyJhbGciOiJSUzI1NiIs...
Step 4 (optional): Automate token refresh with a wrapper script:
#!/bin/bash
# refresh-and-run.sh
export SUPERSET_ACCESS_TOKEN=$(curl -s -X POST \
https://keycloak.example.com/realms/myrealm/protocol/openid-connect/token \
-d "grant_type=client_credentials" \
-d "client_id=superset-client" \
-d "client_secret=$OAUTH_CLIENT_SECRET" | jq -r '.access_token')
export SUPERSET_AUTH_PROVIDER=token
export SUPERSET_URL=http://localhost:8088
node /path/to/mcp-apache/dist/index.js
Then in your MCP config:
{
"mcpServers": {
"superset": {
"command": "/path/to/refresh-and-run.sh"
}
}
}
Option B: Use Superset's login endpoint with OAuth (limited)
Some Superset deployments expose the standard login endpoint even when OAuth is configured (as a fallback). In that case, you can still use database auth for the MCP server by creating a local service account:
# Create a local DB user in Superset even if OAuth is primary auth
superset fab create-admin \
--username mcp-service \
--firstname MCP \
--lastname Service \
--email mcp@internal \
--password STRONG_PASSWORD
Then configure the MCP server with SUPERSET_AUTH_PROVIDER=db and the service account credentials.
LDAP
When Superset uses LDAP authentication:
# superset_config.py
from flask_appbuilder.security.manager import AUTH_LDAP
AUTH_TYPE = AUTH_LDAP
AUTH_LDAP_SERVER = "ldap://ldap.example.com"
AUTH_LDAP_USE_TLS = True
AUTH_LDAP_SEARCH = "ou=users,dc=example,dc=com"
AUTH_LDAP_UID_FIELD = "sAMAccountName"
AUTH_LDAP_BIND_USER = "CN=superset-svc,OU=ServiceAccounts,DC=example,DC=com"
AUTH_LDAP_BIND_PASSWORD = "BIND_PASSWORD"
The MCP server configuration:
SUPERSET_AUTH_PROVIDER=ldap
SUPERSET_USERNAME=your.ldap.user
SUPERSET_PASSWORD=your_ldap_password
Internally this calls the same /api/v1/security/login endpoint but with provider: "ldap".
Access Token (Pre-authenticated)
If you already have a valid Superset JWT (from any method), skip the login flow entirely:
SUPERSET_AUTH_PROVIDER=token
SUPERSET_ACCESS_TOKEN=eyJhbGciOiJIUzI1NiIs...
The server will use this token directly in the Authorization: Bearer header. Note: when the token expires, requests will fail with 401. Use the wrapper script approach above for automatic refresh.
Usage
Claude Code
Add to your project's .claude/settings.json or global ~/.claude/settings.json:
{
"mcpServers": {
"superset": {
"command": "node",
"args": ["C:/development/git/mcp-apache/dist/index.js"],
"env": {
"SUPERSET_URL": "http://localhost:8088",
"SUPERSET_USERNAME": "admin",
"SUPERSET_PASSWORD": "admin"
}
}
}
}
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"superset": {
"command": "node",
"args": ["/absolute/path/to/mcp-apache/dist/index.js"],
"env": {
"SUPERSET_URL": "http://localhost:8088",
"SUPERSET_USERNAME": "admin",
"SUPERSET_PASSWORD": "admin"
}
}
}
}
Development Mode
Run without compiling (uses tsx):
npm run dev
Or in your MCP config:
{
"mcpServers": {
"superset": {
"command": "npx",
"args": ["tsx", "/path/to/mcp-apache/src/index.ts"],
"env": {
"SUPERSET_URL": "http://localhost:8088",
"SUPERSET_USERNAME": "admin",
"SUPERSET_PASSWORD": "admin"
}
}
}
}
Available Tools
Charts
| Tool | Parameters | Description |
|---|---|---|
list_charts |
page?, page_size? |
List all charts with pagination |
get_chart |
id |
Get chart metadata (type, datasource, params) |
get_chart_data |
id |
Execute the chart's query and return result data |
Dashboards
| Tool | Parameters | Description |
|---|---|---|
list_dashboards |
page?, page_size? |
List all dashboards |
get_dashboard |
id |
Get dashboard metadata and layout |
Datasets
| Tool | Parameters | Description |
|---|---|---|
list_datasets |
page?, page_size? |
List registered datasets (tables/views) |
get_dataset |
id |
Get dataset schema, columns, and metrics |
Databases
| Tool | Parameters | Description |
|---|---|---|
list_databases |
(none) | List all configured database connections |
get_database |
id |
Get connection details for a database |
SQL Execution
| Tool | Parameters | Description |
|---|---|---|
execute_sql |
database_id, sql, schema? |
Run an arbitrary SQL query via SQL Lab |
list_saved_queries |
page?, page_size? |
List saved SQL Lab queries |
list_queries |
page?, page_size? |
List recent query execution history |
Running Superset Locally
Docker (quickest)
docker run -d -p 8088:8088 --name superset apache/superset
# First-time initialization
docker exec -it superset superset fab create-admin \
--username admin \
--firstname Admin \
--lastname Admin \
--email admin@example.com \
--password admin
docker exec -it superset superset db upgrade
docker exec -it superset superset init
Access at http://localhost:8088
Docker Compose (full stack with examples)
git clone https://github.com/apache/superset.git
cd superset
docker compose -f docker-compose-non-dev.yml up -d
This starts Superset with Redis, PostgreSQL, and example dashboards pre-loaded.
Architecture
┌─────────────────┐ stdio ┌──────────────────────┐
│ LLM Client │◄──────────────────►│ MCP Server │
│ (Claude Code) │ MCP Protocol │ (this project) │
└─────────────────┘ └──────────┬───────────┘
│ HTTP/REST
▼
┌──────────────────────┐
│ Apache Superset │
│ /api/v1/* │
└──────────┬───────────┘
│
┌──────────▼───────────┐
│ Data Sources │
│ (PostgreSQL, MySQL, │
│ BigQuery, etc.) │
└──────────────────────┘
Auth flow:
- MCP server starts → calls
POST /api/v1/security/login(or uses pre-configured token) - Receives JWT access token + fetches CSRF token
- All subsequent API calls include
Authorization: Bearer <token>+X-CSRFToken - On 401 response → automatic re-authentication and retry
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.