Models
Open-weight models fine-tuned for offensive web security research. Built on Qwen3 and trained on the Strix framework.
kanti-xss-sft-4b
Supervised fine-tuned on curated XSS synthetic Strix traces from real vulnerabilities. Designed to be used as a Strix sub-agent.
kanti-xss-rl-4b
Reinforcement learning variant trained with reward signals from simulated XSS validation. Designed to be used as a Strix sub-agent.
Research & Blog
Write-ups on training methodology, results, and offensive AI security thinking.
Building a Hacker
Details dataset curation and standard fine-tuning for the first iteration of the strix-xss model series.
A Week of RL
The process of creating a RL environment for training and evaluating models for finding XSS vulnerabilities using the Strix Framework
Roadmap
Where we've been and where we're headed.
XSS Foundation Models
Released SFT and RL 4B-parameter models for XSS payload generation. Established training pipeline, dataset curation, and evaluation methodology.
Full Post-Training Run
Completing comprehensive post-training with expanded datasets and improved reward modeling. Next model release coming soon.
Scale & Expand
Scale to larger dense and MoE architectures. Expand beyond XSS to SQLi, SSRF, and other vulnerability classes.
Unified Multi-Vulnerability Model
A single model capable of generating and reasoning about multiple vulnerability types, with integrated defense evaluation.
Fund the Future of AI Security
Kanti Labs is a solo-researcher operation building open-weight offensive security models. Your support — whether as an investor, sponsor, or collaborator — directly accelerates this work.