Let Claude Write SQL Queries Using Your Actual Database Schema

Writing complex SQL is easier when someone already knows your schema. Claude Code reads your migration files, model definitions, and schema dumps, so the queries it writes use your actual table and column names — not guesses.

I need a query that finds all users who placed an order in the last 30 days 
but haven't logged in since. Check my migrations for the correct column names.

Claude reads your database migrations or ORM models, identifies the right tables and columns, and writes a query that works on the first try:

SELECT u.id, u.email, u.last_login_at
FROM users u
INNER JOIN orders o ON o.user_id = u.id
WHERE o.created_at >= NOW() - INTERVAL 30 DAY
  AND (u.last_login_at IS NULL OR u.last_login_at < o.created_at)
GROUP BY u.id, u.email, u.last_login_at;

This works for all levels of query complexity:

# Aggregations
Show me monthly revenue by product category for the last year, 
with a running total column

# Performance investigation
Write a query to find the slowest N+1 pattern — 
which users trigger the most order lookups?

# Data migration
Write an UPDATE query to backfill the new 'full_name' column 
from the existing 'first_name' and 'last_name' columns

# Reporting
Generate a query for a cohort retention table — 
group users by signup month and show what percentage are still active each subsequent month

If you have a database MCP server connected, Claude can go further — running the query, checking the results, and iterating if something looks off. But even without one, reading your schema files is enough to get the structure right.

Describe the data, not the SQL — Claude knows your schema and writes queries that match it.

via Claude Code