
VulnGPT
AI pentesting for business logic & access control
4 followers
AI pentesting for business logic & access control
4 followers
Your app can have secure endpoints and still enforce the wrong rules. VulnGPT is an AI-assisted web app pentesting workspace built to investigate business logic and access control across authenticated users, roles, sessions, and multi-step workflows. It focuses on how an app behaves—not just isolated requests—and helps validate findings with reproducible evidence. Join the private beta and help shape a smarter approach to web security testing.






Free Options
Launch Team

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted

Hey Product Hunt! I'm Abdallah, founder of NovaSecLabs, and I'm building VulnGPT.
I've always found the gap between an application working as intended and an application being secure particularly interesting.
A SaaS application might have authentication, protected endpoints, and all the expected security controls, yet still expose another tenant's data, allow an unauthorized action, or let someone bypass a business workflow.
The difficult part is that these issues often depend on context: which user is acting, which resource they're accessing, what happened earlier, and what the application should allow next.
That's the problem we're building VulnGPT around.
We're developing an AI-assisted pentesting workspace focused on business logic, authenticated workflows, authorization boundaries, and evidence-based validation. The goal is to help developers and security professionals investigate how applications actually behave, while keeping suspected vulnerabilities separate from confirmed findings.
We're early, and we're opening our private beta to people who build, test, and secure real applications.
If you're a founder or CTO, I'd love to hear which application security risks worry you most as your product grows. If you're an engineer or pentester, what are the hardest business logic or authorization issues to reproduce and validate today?
Your feedback would genuinely help shape what we build next.
— Abdallah, NovaSecLabs