SQL Generator Pack

Generate SQL queries from natural language descriptions. Translates plain English into correct, optimized SQL with full schema awareness. Supports PostgreSQL, MySQL, SQLite, and SQL Server dialects.

Quick Start

agentnode install sql-generator-pack

from sql_generator_pack.tool import run

result = run(
    question="Count active users per month",
    schema={"users": ["id", "email", "is_active", "created_at"]},
    dialect="postgresql"
)
print(result["sql"])

Usage

Analytics Queries

result = run(
    question="What is the average order value by product category for the last 30 days?",
    schema={
        "orders": ["id", "total", "created_at"],
        "order_items": ["order_id", "product_id", "quantity", "price"],
        "products": ["id", "name", "category"]
    },
    dialect="postgresql"
)

Data Modification

result = run(
    question="Deactivate all users who haven't logged in since January 2026",
    schema={"users": ["id", "is_active", "last_login_at"]},
    dialect="mysql",
    query_type="dml"
)

Schema Migrations

result = run(
    question="Create a tags table with a many-to-many relationship to posts",
    schema={"posts": ["id", "title", "body", "author_id"]},
    dialect="postgresql",
    query_type="ddl"
)

API Reference

Parameter	Type	Description
question	str	Natural language description of the desired query
schema	dict	Table names mapped to column name lists
dialect	str	"postgresql", "mysql", "sqlite", or "sqlserver"
query_type	str	"auto", "select", "dml", or "ddl"

Returns: sql, query_type, tables_referenced, explanation, is_destructive

License

MIT