Page 1 of 1

AI agent analysis of the source code

Posted: Thu Mar 12, 2026 10:44 am
by oscarlb

Dear VASP staff,

I hope to ask about your stance or license boundaries on making additions or simply analysing the source code of VASP using modern AI cloud agents. The use case I have in mind, for those of us implementing new algorithms or finding bugs in the code, is asking coding agents like chatgpt or claude code to explain unclear parts, direct the user to useful functions existing in the codebase or to find relevant modifications and definitions for the algorithm in question.

As the source has grown quite big over time, this would be a huge help for those who wish to try out focused changes without understanding every part of the code. But I realize the possible privacy issue of the code in question being sent to a third party AI company for processing. As we have entered the age of AI-assisted programming, I thought it prudent to ask.

Best regards,
Oscar


Re: AI agent analysis of the source code

Posted: Thu Apr 09, 2026 8:38 am
by alexey.tal

Dear Oscar,

Sorry for the late reply. Unfortunately, I still don’t have an answer. We are currently reviewing this matter internally and seeking legal advice.
This is obviously a very delicate issue, and we kindly ask for your patience.
I will get back to you within two to three weeks.

Best regards,
Alexey


Re: AI agent analysis of the source code

Posted: Fri Apr 10, 2026 12:04 pm
by alex

Hello Oscar,

one source of information could be your signed license agreement, which probably states that: you are not allowed to distribute the source code to third parties or sth. similar. So as long as you are not running a local copy of your AI model I'd suggest to keep it private.

Best regards,

alex