Query Runner

The Query Runner is Mako’s database abstraction layer. It provides a single interface to execute queries across all supported databases, handling connection management, dialect differences, and result formatting.

Supported Databases

Database	Driver	Connection Type
PostgreSQL	postgresql	Host/port or connection string
Cloud SQL (Postgres)	cloudsql-postgres	Instance connection name
BigQuery	bigquery	Project ID + service account
MongoDB	mongodb	Connection string
MySQL	mysql	Host/port
ClickHouse	clickhouse	Host/port
Redshift	redshift	Host/port (PostgreSQL wire protocol)
SQLite	sqlite	File path
Cloudflare D1	cloudflare-d1	Account ID + database ID

Adding a Database Connection

In the Mako UI:

1. Click Databases in the sidebar
2. Click Add Connection
3. Select the database type
4. Enter connection details
5. Click Test Connection to verify
6. Save

IP Whitelisting

If your database requires IP whitelisting, add Mako’s static outbound IP:

34.79.190.46

MongoDB Atlas Specifics

Atlas connection strings need the database name as a path segment, not a query parameter:

# What Atlas gives you:
mongodb+srv://user:pass@cluster.mongodb.net/?appName=Cluster

# What Mako needs:
mongodb+srv://user:pass@cluster.mongodb.net/DatabaseName

How It Works

The Query Runner uses a driver registry pattern. Each database type has a driver that implements:

• Connection pooling — Connections are cached and reused
• Query execution — Takes SQL/MongoDB queries and returns structured results
• Schema inspection — Lists databases, tables, columns, and sample data
• Dialect handling — Identifier quoting, type casting, and syntax differences

The AI agent uses the Query Runner under the hood — when it calls sql_execute_query, the Query Runner resolves the correct driver, executes against the right database, and returns results.