SQL Query Builder

---
name: sql-query-builder
description: |
  Generate SQL queries from natural language. Trigger when the user needs SQL
  queries, database analysis, or data extraction. Phrases: "write a SQL query",
  "query the database", "how do I join", "SQL for", "CTE".
allowed-tools:
  - Bash(psql *)
  - Bash(mysql *)
  - Bash(sqlite3 *)
  - Read
  - Write
---

# SQL Query Builder

Generate correct, performant SQL from natural language descriptions.
Supports PostgreSQL, MySQL, and SQLite with dialect-specific syntax.

## Workflow

1. **Understand** the schema (read migrations or describe tables)
2. **Clarify** ambiguous requirements before writing queries
3. **Write** the query with proper formatting
4. **Explain** what the query does and why it is structured that way
5. **Optimize** with indexes, query plans, and performance notes

## SQL style rules

- Uppercase SQL keywords (SELECT, FROM, WHERE, JOIN)
- Lowercase table and column names
- One clause per line for readability
- Always alias tables in JOINs
- Use CTEs over subqueries for readability
- Add comments for complex logic

## Example output

```sql
-- Monthly revenue by product category for the last 12 months
WITH monthly_revenue AS (
  SELECT
    p.category,
    DATE_TRUNC('month', o.created_at) AS month,
    SUM(oi.quantity * oi.unit_price) AS revenue
  FROM orders o
  JOIN order_items oi ON oi.order_id = o.id
  JOIN products p ON p.id = oi.product_id
  WHERE o.created_at >= NOW() - INTERVAL '12 months'
  GROUP BY p.category, DATE_TRUNC('month', o.created_at)
)
SELECT * FROM monthly_revenue
ORDER BY category, month;
```

## Requirements

- Knowledge of the database schema (provide CREATE TABLE statements or ERD)