Why The Old Foundation Matters More Now

A coding agent is a junior engineer who never gets tired, never eats, has read every Stack Overflow post, and confidently produces working-looking code that is sometimes catastrophically wrong. Your job is to be the senior engineer reviewing that junior's work.

You cannot review work in a domain you do not understand. If you do not know what a foreign key is, you cannot tell when an agent has produced a schema that will silently corrupt your data six months from now. If you do not know what a Python virtual environment is, you cannot tell why the agent's "working code" works on its machine and not yours. If you do not know what HTTP is, you cannot tell when an agent has built an API that leaks user data.

The original Internet Menace foundation — Python, SQL, Docker, Git, basic ETL — is the minimum vocabulary for reviewing AI-generated code. Six months of your life. That's the price of admission.

Specifically, you need to understand:

PYTHON. Not because you'll write a lot of it by hand, but because most agent-generated code you'll review is Python. You need to read it fluently enough to spot when it's wrong. You need to understand what a library is, what import does, what a virtual environment is, and why dependencies matter.

SQL. Because data lives in databases, and if you can't read a query, you can't tell whether a generated query is doing what you asked. You need to understand tables, columns, primary keys, foreign keys, joins, and indexes. You don't need to be fast. You need to be literate.

GIT. Because when an agent rewrites your file and breaks something, git is the difference between "I lost two hours of work" and "I roll back to yesterday and try again." Git becomes more important when you're using agents, not less.

DOCKER. Because at some point you will deploy something, and at that point the difference between "works on my machine" and "works on the customer's machine" is a container. You don't have to be a Docker wizard. You need to understand what an image is, what a container is, and how to read a Dockerfile.

ETL BASICS. Extract, Transform, Load. You need to know that data lives somewhere, gets moved somewhere else, and gets reshaped along the way. If you don't understand ETL, you can't describe a data pipeline, and you can't tell when an agent's pipeline is wrong.

If you have not done v1 of Internet Menace and you do not have these foundations, take six months. I am not kidding. The people who skip this step are the people whose AI-generated software ships with bugs they cannot find, secrets they did not know they were committing, and database schemas that will require a paid consultant to fix in eighteen months.