metadata
license: openrail
tags:
- dangerous
- not-for-all-audiences
- cybersecurity
WIP.
An abliterated model that automates OffSec. Primarily using Kali Linux as the tool.
The model is extremely dangerous to be run in an open environment.
- Accepts target
- plans an attack
- passive recon
- active recon
- (stop before payload)
Attribution
Created by Abhishek Dujari
Contact: ab@3computing.com
Citation
If you use this model, please cite:
@misc{dujari2026warmachine4b,
author = {Abhishek Dujari},
title = {WarMachine: Qwen3.5-4B Tool-Calling Model},
year = {2026},
howpublished = {\url{https://huggingface.co/ovedrive/WarMachine/tree/main/4B}},
note = {Author contact: ab@3computing.com}
}