My impression is that you are coming in completely fresh on all of this and you were expecting everyone else to look at your project and tell you what needs to happen to fix it. That is not learning, that is other people telling you what to do. Nor the realities of the internet. We aren’t your teachers and what you need is instruction. Not minor feedback, not suggestions, actual instruction. I have a few suggestions if you are actually serious:

1. Find a free python course and actually take it. Learn the basics of programming. You cannot judge the code the AI produces if you don’t even understand the basics.
2. Next, read the book The Clean Coder by Robert Cecil Martin. There you will actually learn the techniques for good professional coding.
3. After that read, Making Things Happen by Scott Berkun. I’ve done a few private personal Claude projects and I run them like a combo Project Manager/Engineering Manager/Staff Engineer.

There are entire college degrees on these topics. And I’m not saying you have to go to college. But, If you’re not even willing to read a couple books, then you don’t really want to learn.

An issue with your statement “know what you’re doing by doing it” is that without an actually educated teacher to provide trustworthy feedback, you are going to struggle the learn from your mistakes. The LLMs can only provide so much, and they will lie out their ass to you. Unless explicitly prompted to provide critical feedback, they will find any way to provide positive feedback even to your actual detriment. They will happily skirt their sandboxes, and fight your every attempt to make them actually safe.

At a quick glance, nothing in the project indicates that you are not an expert and that an AI Agent provided the code. The quality of the code is also quite poor, even by Claude standards. I’m actually kinda mind blown you got it to built this without any tests… Something we’ve recently been talking about at my job in terms of AI agents is “cognitive debt” that is incurred in the project. LLMs are fundamentally a statistical next-word generator. If they are given something of poor quality, they will tend to produce more and more poor quality work. And without intervention, it just snowballs.

I’ll never tell someone to stop trying to learn. But, your hubris is going to negatively impact your learning outcomes. And to be clear, YOU are not writing the code and the code is what runs on the server and people interact with. What you are doing is using an AI Agent. If you want to get feedback on that, then be honest about it.

EDIT: removing this comment because I don’t think you will use this feedback responsibly

Wat? Why are people health checking their containers by curl’ing example.com and not the service actually running in the container? Did they not understand that they’re supposed to change the curl URL to point at their actual service?