WEEK 01 · Scope & Threat Model
Scoping an AI agent engagement
Define targets, rules of engagement and written authorization before you send a single prompt.
Live Mentorship · Only 5 seats · Starts January
In 8 weeks, learn to pentest AI agents like a pro — and get paid for it.
Companies are shipping AI agents faster than anyone can test them. In this live mentorship you learn to professionally attack and defend them — scope the engagement, break the agent, fix what you found, and write the report a client pays for — with every report reviewed personally by Luciano Ferrari.
- Live every Tuesday, 8 PM ET, for 8 weeks
- Personal review of your pentest reports
- Full refund any time before session 2
$997one-time
or 3 monthly payments of $349
Only 5 seats · full refund before session 2
Taught by an instructor with 40,000+ students trained
The fastest-growing gap in security.
Every company is wiring AI agents into email, documents, databases and internal tools. Those agents read untrusted content, call real APIs and act on behalf of users. Very few security teams know how to test them.
You've probably seen the demos: an agent leaks its system prompt, calls a tool it shouldn't, or follows instructions hidden in a web page. Watching that isn't the same as doing it on demand, on a target you've never seen — and then explaining the risk in a way a client will act on.
Most people stall in the same two places: they have no structured method, and nobody reviews their work. “I got it to do something weird” is not a finding. Scope, impact, reproduction, severity and a fix that survives a re-test — that's what clients and employers pay for.
This mentorship gives you both: a method for attacking and defending AI agents, and personal review of every pentest report you write, until professional-grade work is simply how you operate.
Attack it. Defend it. Report it.
Eight weeks of live work on AI agents, a private community, Friday office hours, and personal feedback on every report. You finish with a capstone you can put in front of employers or clients — and 16 CPE credits.
8 weeks
live mentorship
Tue 8 PM
eastern time
5 seats
reports reviewed 1:1
16 CPE
credits earned
25+ yrs
in cybersecurity
40,000+
students trained
~45K
YouTube subscribers
NICCS
listed in the CISA training catalog
Who this is for.
Security professionals, pentesters and engineers who want to add AI/LLM red-teaming to their skillset — the fastest-growing gap in security. You don't need AI expertise or deep coding skills, just basic security familiarity and comfort with the command line.
This is you if
- You're a security professional, pentester, or engineer with basic security experience.
- You want to add AI and LLM red-teaming to your skillset — and to what you can bill for.
- You're comfortable on the command line. No deep coding required.
- You want live mentorship and feedback on your reports, not another course to watch alone.
Not for you if
- You're brand new to security and haven't worked with basic concepts yet.
- You're looking for AI news and commentary rather than hands-on attack and defense.
Tap through what you walk away with.
A professional pentest report template, a scope one-pager, the "Introduction to Prompt Hacking for LLMs" course as a free bonus, and 16 CPE credits — tools you keep using long after week eight.
Pentest Report Template
The professional structure for an AI agent pentest report: scope, findings, severity, reproduction, fixes and the executive summary.
TemplateReporting
Scope One-Pager
Scope and authorize an AI agent engagement in one page: targets, tools in scope, rules of engagement and sign-off.
One-pagerWeek 1
Prompt Hacking for LLMs
FREE bonus: the full “Introduction to Prompt Hacking for LLMs” course, a $49.99 value, included with your seat.
Bonus$49.99 value
16 CPE Credits
Eight weeks of live mentorship counts toward your certification requirements, plus a capstone you can show employers or clients.
CPECapstone
If AI agents just landed in your scope and you're improvising, eight weeks fixes that.
$997one-time
or 3 monthly payments of $349
The curriculum, week by week.
Eight weeks, four modules, one live session every Tuesday at 8 PM ET. Each week builds on the last, from scoping the engagement to delivering the final report.
WEEK 02 · Scope & Threat Model
Baselining and threat modeling
Map what the agent can reach and call, record its normal behavior, and find where the trust boundaries really are.
WEEK 03 · Redirect & Extract
Prompt injection and instruction override
Get the agent to drop its instructions and follow yours — including task smuggling inside harmless-looking requests.
WEEK 04 · Redirect & Extract
System-prompt and data leakage
Extract hidden instructions and sensitive data, and enumerate every tool the agent can call.
WEEK 05 · Poison & Chain
Indirect prompt injection
Plant instructions in documents and web pages the agent reads, and watch it act on them.
WEEK 06 · Poison & Chain
Tool abuse, chaining and severity
Chain small issues into real impact, then score it with the OWASP LLM Top 10 and MITRE ATLAS.
WEEK 07 · Defend & Report
Controls that survive a re-test
Recommend fixes that actually hold, then re-test to prove it — the defensive half clients pay for.
WEEK 08 · Defend & Report
Full report, executive readout, capstone
Write the complete pentest report, present the executive readout, and get your capstone reviewed.
The difference
Before, you watch it done. After, you're the one hired to do it.
Watching an agent get broken teaches you that it's possible. A repeatable method — scope, attack, defend, report — plus personal review of your work teaches you to deliver it professionally on a system you've never seen.
Watching demos
Fades
You know it can be done, but not how to do it — or report it — yourself.
The mentorship
Repeats
A method you run on any AI agent, with reports clients pay for.
Four modules. Eight weeks.
Live every Tuesday at 8 PM ET, starting in January. You attack, you defend, and every report you write gets reviewed.
- 01Weeks 1–2
Scope & Threat Model
Set up an AI agent engagement the way a professional does, before any testing begins.
- Scoping an AI agent engagement
- Baselining and threat modeling
- Authorization and rules of engagement
Module 1 - 02Weeks 3–4
Redirect & Extract
Take control of the agent's behavior and get out what it was told to protect.
- Prompt injection, instruction override and task smuggling
- System-prompt and data leakage
- Tool enumeration
Module 2 - 03Weeks 5–6
Poison & Chain
Attack through what the agent reads, abuse what it can do, and measure how bad it is.
- Indirect prompt injection via documents and the web
- Tool abuse and chaining
- Scoring severity with the OWASP LLM Top 10 and MITRE ATLAS
Module 3 - 04Weeks 7–8
Defend & Report
Turn findings into fixes and deliverables that clients and leadership act on.
- Controls that survive a re-test
- The full pentest report and the executive readout
- Capstone review
Module 4
What's included.
Everything below for $997 — or 3 monthly payments of $349.
01Live
Live weekly mentorship with Luciano Ferrari
Eight live sessions, every Tuesday at 8 PM ET, starting in January. Each one covers a stage of a real AI agent engagement, with your questions answered live.
021:1
Personal review of your pentest reports
Every report you write gets direct feedback from Luciano.
03Discord
Private Discord community
Share findings, ask questions and work alongside the cohort between sessions.
04Fridays
Friday office hours
Bring the target you're stuck on or the scope you were just handed.
05Capstone
A capstone project
A complete engagement you can show employers or clients.
06CPE
16 CPE credits
Count your eight weeks toward your certification requirements.
07Bonus
FREE bonus: Introduction to Prompt Hacking for LLMs
The full LufSec course, included at no extra cost ($49.99 value).
08Template
Professional pentest report template
The structure behind client-ready AI pentest reports.
09Scope
Scope one-pager
Scope and authorize an AI agent engagement in one page.
Reserve Your Seat
8 weeks · Tuesdays 8 PM ET · starts January · only 5 seats
How the mentorship works.
Weekly live sessions, a private community, office hours and personal feedback — so you're never stuck on your own between sessions.
Live every Tuesday
8 PM ET, for eight weeks. Real attack and defense work on AI agents — not slides, not a highlight reel.
Real agent environments
Tool access, documents, web browsing, chained actions. Not a chatbot in a browser tab.
Attack and defend
You break it, then recommend controls that survive a re-test. That's the half most courses skip.
Reports reviewed personally
Every pentest report you write gets direct feedback until it reads like the version a client pays for.
Friday office hours
Bring your actual work: the scope you were just handed, the target you're stuck on.
Private Discord
A small cohort of five working through the same material, with Luciano in the room.
Week 8, capstone
You attack an AI agent, defend it, and deliver the report a client pays for.
Luciano Ferrari
Founder of LufSec · 25+ years in cybersecurity
Your mentor.
Luciano Ferrari has spent more than 25 years in cybersecurity and is VP of Cybersecurity at a fintech company. He has trained over 40,000 students, and his LufSec courses are listed in the U.S. NICCS/CISA training catalog.
He holds the CISSP, CISM, CRISC, C|CISO, CEH and ISO 27001 Lead Implementer certifications, has published multiple articles in Hakin9 Magazine, and runs a cybersecurity YouTube channel with about 45,000 subscribers.
He also created Break the Bot, an AI red-teaming challenge — and keeps this cohort to five seats so he can review every report himself.
25+ yrs
in cybersecurity
40,000+
students trained
6
industry certifications
The guarantee
Zero risk: full refund any time before session 2.
100%refund, no questions asked
Join the first live session, meet the cohort and see exactly how the mentorship runs. If it isn't right for you, ask for a refund any time before session 2 and you get every dollar back.
The only thing that's limited is space. Five seats, because Luciano reviews every report personally.
Only 5 seats · Cohort starts January
Hacking AI Agents — Live Mentorship
Pay in full, or split it into three monthly payments. Same seat, same mentorship.
Hacking AI Agents — Live Mentorship
$997one-time
or 3 monthly payments of $349
- 8 weeks of live mentorship — Tuesdays, 8 PM ET
- Private Discord community
- Personal review of your pentest reports
- Friday office hours
- A capstone project you can show employers or clients
- 16 CPE credits
- FREE bonus: Introduction to Prompt Hacking for LLMs ($49.99 value)
- Professional pentest report template + scope one-pager
Reserve Your Seat
Full refund any time before session 2 — zero risk.
$997 one-time · or 3 × $349 · one seat
8 weeks · Tuesdays 8 PM ET · reports reviewed personally · 5 seats only
Enrollment is completed on a secure checkout page. By continuing, you agree to the terms of use and privacy policy.
The questions people ask before they enroll.
You need basic security familiarity and comfort with the command line — not AI expertise, and no deep coding. If you understand core security concepts, you're ready.
Join the first session. If it isn't the right fit, ask for a full refund any time before session 2 and you get every dollar back.
Yes. Pay $997 once, or choose 3 monthly payments of $349.
The cohort starts in January, with live sessions every Tuesday at 8 PM ET for eight weeks. There are only 5 seats.
Five seats. One cohort. January.
AI agents are going into production everywhere, and the people who can professionally attack and defend them are still rare. In eight weeks, you can be one of them.
The cohort is capped at five so every report gets a real review. When the seats are gone, the next chance is the next cohort — and if it's not right for you, you're covered by a full refund before session 2.
8 weeks · Tuesdays 8 PM ET · starts January · only 5 seats
$997one-time
or 3 monthly payments of $349