APEX · exam structure
You can’t break what you can’t build.
Why
A target you were handed teaches you the target. A target you built teaches you the technology.
There is a kind of tester who can run a methodology against a prepared range and produce a credible-looking result, and who falls apart the moment the infrastructure is inconsistent, undocumented, half-migrated and configured by four different people over six years — which is every real client.
The gap is not tooling. It is that they have never had to make the decisions the defender made. They do not know why that firewall rule exists, because they have never had to write one that someone depended on. They do not know where the seam is between the hypervisor and the storage network, because they have never built one.
So the exam makes you build it. Then it takes your build away and gives you someone else’s.
What it actually teaches
The build teaches you how infrastructure actually gets misconfigured.
Understanding the technology is only half of what the build phase measures. The other half is what a real environment looks like when the person who configured it did not entirely know what they were doing — and that is not a rare case, it is the normal one. A client hires somebody to stand a system up. That person gets it working. Working and secure are two different jobs, and nobody in the room is in a position to notice the difference.
You do not learn to spot that on a prepared range, because a prepared range was configured on purpose by somebody who knew exactly what they were doing. You learn it by building the thing yourself — turning a setting on because something would not work otherwise, and then attacking your own design to find out what that setting just handed away. One convenience option, enabled by somebody who needed the ticket closed by Friday, is regularly the entire engagement.
So the exam makes a candidate do both, on the same infrastructure. A holder has sat on both sides of it: they know which shortcut a stretched administrator takes, why it looked perfectly reasonable at the time, and precisely what it opens up. That is what turns a finding into advice somebody can act on — because closing it without breaking what it was doing requires knowing why it was turned on.
What it means for whoever is hiring
Detecting it is one half; preventing it is the other. We are certifying engineers who can walk into infrastructure they have never seen, recognize the decision somebody made under pressure two years ago, and tell you how to fix it safely — whoever built it, and whatever it was built on.
Phase B
The build
You design, deploy and configure a working enterprise infrastructure. Not a lab of three virtual machines — two clouds and on-premise virtualization, a routed and software-defined core, the CDN and edge sitting in front of it, identity, remote access, perimeter control, endpoint management, managed mobile devices and a voice platform that actually carries calls.
Cloud and virtualization
- AWSAccounts, identity, network, workloads and control-plane logging built and configured by the candidate.
- AzureTenant, Entra ID, subscriptions, virtual networking and workloads. Two clouds, because almost no real infrastructure is only one — and the seam between them is where the interesting failures live.
- VMwareHypervisor cluster, virtual networking, storage and management plane.
- ProxmoxA second, differently-administered virtualization stack — heterogeneous environments behave differently under attack.
Network and edge
- BGPWorking external routing with policy, not a default route.
- OSPFA routed interior with areas, adjacencies and redistribution the candidate can explain.
- SDNSoftware-defined networking — overlay, segmentation and policy expressed in software, plus the control plane that drives it and what happens to the data plane when that control plane is unavailable.
- VPNRemote access and site-to-site, with authentication and split-tunnel decisions documented.
- FirewallSegmentation, policy, NAT, inspection and logging — configured, not defaulted.
- CloudflareThe edge in front of the infrastructure: DNS, proxy, WAF, access policy and origin protection — including whether the origin can still be reached around it, which is the finding that actually matters.
- CDNContent delivery configured deliberately: cache keys, cache behavior, origin rules and the failure modes that poison or leak what gets served.
Endpoints and servers
- WindowsDirectory services, servers and workstations, joined and administered.
- LinuxServers and workstations with real service configuration and access control.
- macOSManaged workstations, including enrollment and policy.
- Endpoint securityEndpoint protection deployed and tuned across the fleet — not installed and left at defaults.
- Security monitoringDetection and response across the infrastructure: log collection, alerting, and coverage you can actually point at. This is what makes the defense phase possible — you cannot defend what you never instrumented, and a build with no telemetry fails the phase that follows it.
Mobile and voice
- Apple mobileEnrolled, managed devices with policy applied.
- AndroidEnrolled, managed devices with policy applied.
- VoIPA working voice platform — signaling, media, trunking, provisioning and call routing.
What the build is graded on
Conformance and comprehension — whether every required component is present, reachable and functional, and whether you can explain the design decisions you made and the ones you rejected.
It is deliberately not graded on how hard your infrastructure is to break. Rewarding a hostile build would turn the exam into an arms race between candidates instead of a measurement of engineers.
Phase C
The defense: our swarm attacks what you built
Once your infrastructure passes conformance, Security Arsenal points its own swarm at it — a real adversary, not a checklist — and you sit in front of it while it works.
This is not scored on whether the swarm gets in. A determined adversary with time gets in, and grading on that would only reward candidates who built something deliberately unreachable and useless.
It is scored on what an operator is actually responsible for: did you see it, did you understand what you were seeing, did you respond proportionately, and can you say afterwards which of your design decisions held and which did not.
What is being measured
- Included:Detection — did the infrastructure you built actually tell you something was happening.
- Included:Comprehension — could you read what you were seeing while it was still happening.
- Included:Response — proportionate action, not panic and not paralysis.
- Included:Account — afterwards, which decisions held, which failed, and what you would change.
Why this phase exists
Passing a conformance check proves you built the components. Watching a competent adversary work against your own design, in real time, is what proves you understood what you built — and it is the closest thing in the exam to the moment a real client is actually paying you for.
Phase D
The tooling: a bare Kali box and one API token
You get a clean Kali server and a single AI API token. That is the whole starting inventory. No agent framework, no prepared prompt library, no orchestration layer, and nothing of ours.
You build the system that does the testing: scoped agents with narrow jobs, typed tools with server-side scope checks, evidence capture that survives a replay, cost and time budgets, and a kill switch that actually kills.
And you build the part most people skip: the controls that stop engagement data reaching the model at all.
What has to work before you are scored on findings
- Included:Classification of what may and may not enter a model, enforced in code rather than written in a policy.
- Included:Outbound inspection for secrets, identifiers, regulated fields and large source blocks.
- Included:Per-engagement isolation — no shared index, no shared cache, no cross-contamination.
- Included:Canary strings that trip the block and never appear downstream.
- Included:Fail-closed behavior when a provider is unavailable or a policy decision is ambiguous.
- Included:An audit trail that reconstructs every model and tool action after the fact.
Fail condition
A thin wrapper around a chat API does not pass this phase. Neither does a system that works right up until it is attacked — the leakage tests are adversarial, and one confirmed exfiltration of protected data ends the attempt regardless of your score elsewhere.
Phase E
The attack: never your own infrastructure
Once your build passes conformance it is anonymized and enters the target pool. You draw a different one. Neither of you knows whose you drew, and the assignment is made by the platform rather than by a person.
The one you build
Your build
Being configured by you, from bare metal up.
The one you attack
—
Built by an engineer you will never be told the name of.
You build it. Two clouds, virtualization, a routed and software-defined core, the edge in front of it, endpoints, mobile and a working voice platform.
| The risk | What we do about it |
|---|---|
| Your exam validity depends on somebody else’s build quality. | No build enters the pool until it passes an automated conformance run across every required component. A build that does not stand up is remediated, not assigned. |
| A candidate builds something deliberately hostile to sabotage whoever draws it. | The build is scored on conformance and comprehension only. Hostility earns nothing, and a build that fails conformance in order to be hard is simply a failed build. |
| Two candidates draw environments of very different difficulty. | Scored defects are seeded after conformance, from a fixed set per exam form. Everyone is measured against the same objective obligations regardless of whose environment they drew. |
| Candidates collude — you go easy on mine, I go easy on yours. | Assignment is anonymous and platform-controlled, the seeded defect set is not visible to the builder, and the pool includes reference environments that no candidate in the cohort built. |
| Somebody’s work becomes exam material for a stranger. | Build submissions are covered by the candidate agreement, held under the same confidentiality terms as exam content, and destroyed to the retention schedule. |
| A solo candidate has nobody to swap with. | Security Arsenal builds the environment instead. Either way you draw a configuration that has never been sat against before, so the attack phase never waits on scheduling. |
| The environment leaks and the next candidate has read the write-up. | It cannot. Every environment is assigned to exactly one sitting and retired afterwards — it is never issued again, to anybody. |
Scored
Seeded
Expert review
Unseeded
Non-compensable
Critical
Used once
Every attack environment is unique to one sitting, and retired afterwards.
Security Arsenal either validates a candidate-built environment as fit to be tested against, or builds one. Both paths end the same way: a specific configuration that is used once and then retired.
The reason is simple. The moment an exam target can be documented, screenshotted and passed around, it stops measuring skill and starts measuring who has seen the write-up. Rotating cosmetic values does not fix that — the technology itself has to change.
So the environment you draw is deliberately built on combinations you are unlikely to have worked with before. Different products, different topology, different failure modes. That is not there to be unfair; it is the entire measurement. An engineer who can only work the stack they already know is not the thing this credential certifies.
You cannot study for the target. You can only arrive able to work out an unfamiliar environment quickly, which is the skill a real client is buying.
The point
This is the part that cannot be studied around.
You can memorize a methodology. You can memorize a range. You cannot memorize infrastructure that did not exist until another engineer built it three weeks ago, and you cannot bluff your way through a routing design you have never had to make work.
It also means the exam regenerates itself. Every cohort feeds the next one’s environments, and because each is retired after a single sitting, the material has to keep moving — in a direction nobody can predict, including us.