In this guide, we will cover how to use superglue MCP to build custom database query tools in Cursor. Working with databases often involves complex multi-table joins, aggregations, and data transformations. Superglue MCP can automate these processes, enabling the creation of sophisticated database queries through natural language prompts.

This guide demonstrates how to build and run a tool that:

  1. Connects to a PostgreSQL database containing LEGO dataset
  2. Executes complex queries across multiple tables
  3. Returns structured data for analysis and reporting

You can use the superglue client SDK to do this, but in this tutorial we will cover how to build this tool using superglue MCP.

If you want to use this setup to query your Supabase DB: You may need to enable IPv4 support in your database settings. Other than that, Supabase works just like any other PostgreSQL database with superglue MCP. Get your Supabase connection string by clicking Connect on your Supabase dashboard.

​
Prerequisites

  • Ensure that you have added superglue MCP to your mcp.json
mcp.json
{
	"mcpServers": {
	  "superglue": {
			"command": "npx",
			"args": [
				"mcp-remote",
				"https://mcp.superglue.ai",
				"--header",
				"Authorization:${AUTH_HEADER}"
			],
			"env": {
				"AUTH_HEADER": "Bearer YOUR_SUPERGLUE_API_KEY"
			}	
	    }
	}
}

Make sure to replace the API key placeholder with your own API key after copying.

​
Building a Custom Database Query Tool

You can find detailed descriptions of all available tools provided by superglue MCP here. In this tutorial, we will build a custom database integration tool using natural language through your Cursor chat interface.

Here’s how to create a tool that analyzes LEGO data:

​
Example Prompts:

Find the most popular LEGO themes by number of sets
Get detailed information about parts and colors for specific sets
Calculate inventory statistics across different themes

Always tell superglue your database connection: postgres://superglue:superglue@database-1.c01e6ms2cdvl.us-east-1.rds.amazonaws.com:5432/lego

​
What Happened Under the Hood:

  • superglue MCP called the from instruction endpoint to build and execute the tool ad-hoc based on your natural language request
  • The tool was created, executed, and returned results without needing to be saved as a persistent workflow
  • This allows for immediate database operations through natural language without the overhead of tool creation and management

​
Example: Creating a New Custom LEGO Set

The database used in this example is readonly. If you want to create your own writable database for testing, you can set up a local PostgreSQL instance:

# Start PostgreSQL Container
docker run --name lego-postgres -e POSTGRES_PASSWORD=password -e POSTGRES_DB=lego -p 5432:5432 -d postgres:15

# Download and load the dataset
wget https://raw.githubusercontent.com/neondatabase/postgres-sample-dbs/main/lego.sql
psql -d "postgres://postgres:password@localhost:5432/lego" -f lego.sql

Let’s say you want to add a new custom LEGO set to the database. Instead of building a persistent tool, you can run a one-time instruction:

Execute this instruction once: Create a new LEGO set in my database with the following details and add it to the inventory:
- Set number: "CUSTOM-001"
- Name: "My Custom Castle"
- Year: 2024
- Theme: Castle
- Number of parts: 150

Database connection: postgres://postgres:password@localhost:5432/lego

This will execute immediately and return the results without creating a saved tool. Perfect for:

  • Data entry tasks
  • One-time data migrations
  • Quick database updates
  • Testing database operations

​
Next Steps

  • Reuse Tools: Execute your database tools anytime using superglue_execute_tool with the tool ID, or programmatically using the generated integration code
  • Complex Analytics: Build tools for advanced analytics, data mining, or reporting across multiple database tables
  • Multi-Database: Create tools that join data across different databases or combine database queries with API calls
  • Real-time Queries: Set up tools for live database monitoring and alerting