Large Language Models have transformed how security professionals develop custom tooling. But asking an AI for "the whole application" often results in hallucinated APIs, inconsistent code, and tools you don't understand. This hands-on workshop introduces "vibe coding" - the practice of using AI to rapidly prototype security tools through iterative, grounded prompting.
Attendees will learn the "think then act" methodology: first using AI to create a detailed Product Requirements Document (PRD), then implementing features phase-by-phase with verification at each step. We'll apply this approach to build a functional shellcode loader generator with configurable evasion techniques.
By the end of this workshop, participants will have a working Dockerized web application that generates custom C# payloads with options including payload bloating, sleep delays, anti-sandbox checks, environmental keying, and entropy reduction - all created through AI-assisted development.
Who Should Attend?
Red team operators
Penetration testers
Security professionals
Blue team members
Anyone interested
Learning Objectives:
Apply the "think then act" methodology for AI-assisted security tool development
Write effective prompts that minimize LLM hallucinations using explicit constraints and concrete examples
Create a Product Requirements Document (PRD) for custom security tooling using AI
Implement common evasion techniques: payload bloating, sleep delays, anti-sandbox checks, and environmental keying
Understand the difference between in-process shellcode execution and process injection techniques
Jean-François is based in Portugal, where he is the CEO of Offensive Guardian, a boutique red and purple teaming shop providing freelance services to various organizations. He has worked for other noteworthy firms, including, but not limited to: Neuvik, TrustedSec, Fortra's Cobalt-Strike team, and NVISO.
