Describe Your Data Needs in English and Let Claude Write the SQL

Complex SQL is hard to get right. Describing what you want and having Claude write it — with your actual table names and relationships — is faster and less error-prone.

"I need monthly revenue by product category, excluding refunded orders, for Q4 2024"

Claude reads your database schema (through migrations, models, or an MCP database connection), then writes a query using your actual table and column names:

SELECT 
    c.name AS category,
    DATE_TRUNC('month', o.created_at) AS month,
    SUM(oi.price * oi.quantity) AS revenue
FROM order_items oi
JOIN orders o ON o.id = oi.order_id
JOIN products p ON p.id = oi.product_id
JOIN categories c ON c.id = p.category_id
WHERE o.created_at BETWEEN '2024-10-01' AND '2024-12-31'
AND o.status != 'refunded'
GROUP BY c.name, DATE_TRUNC('month', o.created_at)
ORDER BY month, revenue DESC;

This works for any complexity level:

# Simple lookups
"Find all users who signed up in the last 7 days but haven't verified their email"

# Complex aggregations
"Show the top 10 customers by lifetime value with their order count and average order size"

# Window functions
"Rank products by sales within each category and show the running total"

# Data investigation
"Find orders where the total doesn't match the sum of line items"

If you're using an ORM, ask Claude to write it in your framework's query builder instead:

"Write that as an Eloquent query, not raw SQL"
"Give me the Prisma version"
"Convert this to a Django ORM queryset"

Claude can also optimize existing slow queries — paste the SQL and your explain plan, and it rewrites the query or suggests indexes.

SQL is a language for machines. English is a language for humans — describe the data, let Claude write the query.

via Claude Code