Interested in understanding of how AI-powered Subdomain Mutations work

[Greatz08]

First of all i would like to thank you for this awesome open source project dev :-)).
I am interested in understanding about this feature which is mentioned in repo main page - "AI-powered Subdomain Mutations"
What it is and how it works and what is ai thing in this and overall working of it.
Thanks in advance x)

[TheTechromancer]

Thanks for your interest in the subdomain mutations. I should start by saying that our method for this is a very simple implementation of NLP, but it's both efficient and effective.

We don't use large language models or anything so sophisticated (although I plan to experiment with this in the future). Instead, over the course of a BBOT scan, every discovered DNS name is analyzed and converted into mutations, which are accumulated into a "mutator" that counts the occurrences of each one.

This analysis is key. We use the wordninja library, trained on a DNS-specific dataset, to split the hostname as many ways as possible, including intelligent splitting by word. The code for this is here.

Finally at the end of the scan, we take the most commonly occurring mutations and apply them recursively to all subdomains found by the scan.