A real application, written fast with AI. You find what it left open.
The group starts with nothing: no account, no documentation, no source code. Just an application that looks perfectly ordinary and that nobody reviewed before it went live. Your job is to say where you would try to get in.
"It works, ship it"
An internal application, built over a weekend with whatever the AI suggested, copy-pasted and pushed straight out. Security was going to be next sprint. Its author has since left the company, and it went to production on a Monday.
Your committee picks it up the way an attacker would: start from nothing, call out angles out loud, and see how far they go. An engineer holds the keyboard and runs every idea, including the ones you assumed were naive. They go far.
The group's objective for the session: complete control of the server.
Gain complete control of the server
- 1See what a stranger sees
- 2Get inside the application
- 3Walk out with the customer database
- 4Run your own commands on the server
- 5Own the whole thing
Five steps, five things worth understanding
Each level exposes a new surface and teaches the same lesson: small things, handled early, break the entire chain.
-
Door 1 · Seen from outside
See what a stranger sees
Before any account or password: what your company leaves visible from the open internet. The group says where it would look, the screen shows what is there.
-
Door 2 · Getting in
Get inside the application
Someone suggests lying to the login form. Three characters later the session belongs to someone else, then to the administrator. No password involved.
-
Door 3 · The data
Walk out with the customer database
The group asks for the customer list and it appears. The passwords fall within the minute. This is the slide in every breach post-mortem, except it unfolds live.
-
Door 4 · The server
Run your own commands on the server
An idea from the group makes the machine underneath run a command. From here it stops being an application problem and becomes a company problem.
-
Door 5 · Total control
Own the whole thing
One permission nobody tightened, and the entire server falls. Every file, every customer, every secret, and still no alarm raised.
The weaknesses you will actually use
Each one maps to the OWASP Top 10, the industry reference. Nobody is going to tell you about them. You are going to use them.
| OWASP | Weakness | In practice |
|---|---|---|
| A03 | Your database, extracted | A search box that trusts whatever it is handed. Known as SQL injection, and still the most common way customer data leaves a company. |
| A02 | Passwords cracked in seconds | Passwords stored with obsolete protection, secrets committed straight into the code. Breaking them needs a laptop and a few minutes. |
| A01 | Reading other customers' files | Change one number in the address bar and you are looking at someone else's data. Then quietly promote yourself to administrator. |
| A03 | The server runs your commands | Input that travels all the way to the operating system. This is the exact moment an application incident becomes a company incident. |
| A04 | A file that opens a door | An upload form that accepts more than it should, and hands over control of the machine that runs it. |
| A05 | From a foothold to everything | A single over-permissive setting turns limited access into complete ownership of the server. |
| A03 | Code running where it never should | A page that evaluates whatever a visitor types into it. Another route to executing code on your own systems. |
| A08 | Trusting data you cannot trust | The application rebuilds internal objects from outside input, and its integrity is gone. |
What your committee can do afterwards
None of it is technical. These are the SecOps mechanics of a modern startup, seen from the inside, once and for all.
-
Why security levels down, not up
You will watch an application that is sound almost everywhere fall because of one neglected spot. That mechanism never appears on a compliance dashboard.
-
How a detail becomes a company-wide incident
None of the weaknesses found is serious on its own. Chained together they hand over the server. That is always how it happens.
-
What shipping speed actually costs
Shipping fast is a competitive advantage, not a fault. You will see exactly where that choice leaves marks, so you can arbitrate where it matters instead of slowing everything down.
-
The questions to put to your teams
A handful of simple questions separate a team that owns its exposure from one that does not. You leave with those questions, and with the answers that should satisfy you.
-
What an incident sets off beyond the technical
72-hour notification, customers to inform, due diligence reopened, valuation renegotiated. You watch the leak happen, then we walk through what it triggers.
It is almost always something small
An input nobody validated, a password stored carelessly, a permission left wide open, a limit never set. None of it is sophisticated, and that is exactly the point: the fixes were cheap and early, and nobody prioritised them.
Send your teams through it next
You leave with the application's source code and every flaw documented. Your engineering teams can run exactly the same chain, so the conversation continues on a shared reference instead of a memo.
Shall we build this for your committee?
Framed around your sector, your systems and your real exposure.
Reply within 48 business hours · no commitment