---

name: sql description: "Database querying and analysis using SQLAlchemy 2.0+ with support for PostgreSQL, MySQL, SQLite, and SQL Server. Use when tasks require: (1) Querying databases via SQL, (2) Reading data into DataFrames for analysis, (3) Performing database operations with proper transaction handling. Environment variable with connection string must be set (check resources/RESOURCES.md for available databases and schemas)." license: "© 2025 Daisyloop Technologies Inc. See LICENSE.txt"

SQL Database Integration

Before You Start

Read resources/RESOURCES.md first to understand available databases, table structures, column types, and relationships before writing any queries. Alternatively, run scripts/inspect_schema.py to discover schema programmatically.

Overview

Query databases using SQLAlchemy 2.0+ with pandas integration for analysis. Supports PostgreSQL, MySQL, SQLite, and SQL Server.

Quick Start

Copyimport os
import pandas as pd
from sqlalchemy import create_engine, text

# Connection string from environment (set via resources)
engine = create_engine(os.environ["DATABASE_URL"])

# Read query results into DataFrame
df = pd.read_sql("SELECT * FROM users LIMIT 10", engine)
print(df.head())

Connection Patterns

Using Environment Variables

Connection strings are stored as secret resources and injected as environment variables. Check resources/RESOURCES.md for the variable name.

Copyimport os
from sqlalchemy import create_engine

# Get connection string from environment
conn_str = os.environ["DATABASE_URL"]  # or your specific env var name
engine = create_engine(conn_str)

Engine-Specific Patterns

See references/engines.md for connection string formats and driver-specific details for:

• PostgreSQL (psycopg)
• MySQL (pymysql)
• SQLite (built-in)
• SQL Server (pyodbc)

Read Operations

Basic Queries with pandas

Copyimport pandas as pd
from sqlalchemy import create_engine

engine = create_engine(os.environ["DATABASE_URL"])

# Simple query
df = pd.read_sql("SELECT * FROM orders WHERE status = 'pending'", engine)

# With parameters (safe from SQL injection)
df = pd.read_sql(
    "SELECT * FROM orders WHERE customer_id = :cid",
    engine,
    params={"cid": 123}
)

Using SQLAlchemy Core

Copyfrom sqlalchemy import create_engine, text

engine = create_engine(os.environ["DATABASE_URL"])

with engine.connect() as conn:
    result = conn.execute(text("SELECT * FROM products WHERE price > :min_price"), {"min_price": 100})
    for row in result:
        print(row)

Write Operations

⚠️ EXTREME CAUTION REQUIRED

Write operations (INSERT, UPDATE, DELETE) cannot be undone. Always:

1. Use --dry-run first to preview changes without committing
2. Verify the WHERE clause - a missing or wrong WHERE affects all rows
3. Start with SELECT - run the equivalent SELECT to see affected rows
4. Back up data before bulk updates on production

Copy# ALWAYS dry-run first
uv run scripts/execute_sql.py "UPDATE users SET active = false WHERE id = 123" --dry-run

Transactions

Copyfrom sqlalchemy import create_engine, text

engine = create_engine(os.environ["DATABASE_URL"])

# Transaction with automatic commit/rollback
with engine.begin() as conn:
    conn.execute(text("UPDATE orders SET status = :status WHERE id = :id"), {"status": "shipped", "id": 123})
    conn.execute(text("INSERT INTO order_history (order_id, event) VALUES (:oid, :event)"), {"oid": 123, "event": "shipped"})
# Commits automatically if no exception, rolls back on error

Insert with Returning

Copywith engine.begin() as conn:
    result = conn.execute(
        text("INSERT INTO users (name, email) VALUES (:name, :email) RETURNING id"),
        {"name": "Alice", "email": "alice@example.com"}
    )
    new_id = result.scalar()
    print(f"Created user with ID: {new_id}")

Further Analysis with Pandas

After loading data, use pandas for analysis:

Copyimport pandas as pd
from sqlalchemy import create_engine

engine = create_engine(os.environ["DATABASE_URL"])

# Load data
df = pd.read_sql("SELECT * FROM sales", engine)

# Analysis
summary = df.groupby("region").agg({
    "revenue": ["sum", "mean"],
    "quantity": "sum"
})

# Filter and transform
recent = df[df["date"] >= "2024-01-01"]
recent["margin"] = recent["revenue"] - recent["cost"]

# Export results
summary.to_excel("output/sales_summary.xlsx")

Large Datasets with Chunking

Copy# Process large tables in chunks
chunks = pd.read_sql("SELECT * FROM large_table", engine, chunksize=10000)

for chunk in chunks:
    # Process each chunk
    processed = chunk[chunk["status"] == "active"]
    # ... further processing

Best Practices

1. Always parameterize queries - Never use f-strings or string concatenation for user inputs
2. Use transactions for writes - Ensures atomicity and enables rollback
3. Check schema first - Read resources/RESOURCES.md before writing queries
4. Close connections - Use context managers (with) to ensure proper cleanup
5. Limit result sets - Use LIMIT during development to avoid pulling large datasets

Scripts

Ready-to-use scripts for common database operations:

• scripts/inspect_schema.py - Discover tables, columns, keys, and indexes
• scripts/query_to_dataframe.py - Execute queries and export as CSV/JSON
• scripts/execute_sql.py - Run write operations with transaction safety and dry-run

Copy# Examples
uv run scripts/inspect_schema.py --table users
uv run scripts/query_to_dataframe.py "SELECT * FROM orders" --format csv
uv run scripts/execute_sql.py "UPDATE users SET active = true" --dry-run

References

• references/engines.md - Connection strings, SQL dialect differences, and gotchas
• references/patterns.md - Advanced SQL patterns:
  • Window functions (ROW_NUMBER, LAG/LEAD, running totals)
  • CTEs (recursive and non-recursive)
  • Pivoting data
  • Date/time operations by engine
  • JSON operations by engine