How I Review AI-Generated Python Code

How to save time with AI while improving code quality.

No matter what, you'll always be writing custom monitoring code for cloud services despite the number of monitoring features AWS provides. Recently, I had the pleasure of writing a service in response to an unforeseen outage. I loaded up PyCharm (VIM users, please spare me) and navigated to my good friend ChatGPT to hash this out.

Part of this monitoring job required a call to retrieve a list of ECS services based on the cluster name, and I prefixed the names of the services for filtering.

With this information, I started with a simple prompt for ChatGPT to get the ball rolling.

Write code to match a prefix to a list of services on an ECS cluster

It gave a lovely function that worked out of the box.

import boto3
from botocore.exceptions import BotoCoreError, ClientError

def match_services_with_prefix(cluster_name, prefix):
    ecs_client = boto3.client("ecs")
    matched_services = []
    next_token = None

    try:
        while True:
            response = ecs_client.list_services(
                cluster=cluster_name,
                nextToken=next_token
            )

            service_arns = response.get("serviceArns", [])

            # Extract service…