Claude Code Generation and Review

Claude excels at generating and reviewing code, making it a productivity multiplier for ShopMax India's engineering teams. From writing SQL queries for sales analytics, to generating FastAPI endpoint boilerplate, to reviewing pull requests for security issues, Claude accelerates development while catching bugs before they reach production. Unlike simple autocomplete, Claude understands business context - you can describe what the code should accomplish and Claude produces idiomatic, production-ready code with error handling.

For code generation, provide Claude with clear requirements including input/output types, error handling expectations, and any framework or style conventions. For code review, send the code with specific review criteria - security, performance, readability, or correctness. Claude returns actionable feedback with specific line references and suggested fixes. For automated review pipelines, use structured output to get machine-parseable review results with severity levels.

The following example shows ShopMax India using Claude to generate a database query function and then review it for security and performance issues:

It gives the following output,

=== GENERATED CODE ===
import psycopg2
from typing import Optional

def get_top_products(city: str, category: str, limit: int = 10) -> list:
    try:
        conn = psycopg2.connect(dsn=os.getenv("DB_DSN"))
        with conn.cursor() as cur:
            cur.execute(
                "SELECT product_id, name, SUM(units) AS units_sold, "
                "SUM(revenue) AS revenue_rs FROM sales "
                "WHERE city = %s AND category = %s "
                "GROUP BY product_id, name ORDER BY units_sold DESC LIMIT %s",
                (city, category, limit)
            )
            rows = cur.fetchall()
        return [{"product_id": r[0], "name": r[1],
                 "units_sold": r[2], "revenue_rs": float(r[3])} for r in rows]
    except psycopg2.Error as e:
        print(f"DB error: {e}")
        return []
    finally:
        if conn:
            conn.close()

=== CODE REVIEW ===
LGTM - The code is well-structured with parameterized queries (no SQL injection risk),
proper error handling, and resource cleanup via finally block.
Medium: Line 5 - Connection created inside function on every call. Consider using
a connection pool (psycopg2.pool.SimpleConnectionPool) for production to avoid
connection overhead on high-traffic ShopMax India dashboards.

For ShopMax India engineering teams, integrate Claude code review into your GitHub Actions CI pipeline - trigger a review on every pull request and post the findings as a PR comment. Use Sonnet for code generation (fast and cost-effective) and Opus for security-focused reviews of critical code like payment processing and customer PII handling. Provide Claude with your team's coding standards document as context to get reviews aligned with your conventions rather than generic best practices.