How CyberForge Academy works
There are no human instructors. A supervisor agent routes every request through safety review to a specialist, reviews the output, and records the decision — so training stays personalized, transparent, and lab-only.
24
AI agents on staff
10
Courses
10
Guided labs
15
CTF challenges
The 10-step orchestration workflow
Every interaction follows the same auditable pipeline, coordinated by the supervisor agent Axis.
- 1
Request received
Every message, lab action, quiz submission, or report you send arrives at the supervisor agent, Axis.
- 2
Intent classified
Axis passes the request to the Intent Classifier (Prism), which determines what you are actually asking for and which specialist owns it.
- 3
Safety review
The Safety Guardian screens the intent for anything that would target real systems, violate authorization, or cross an ethical boundary.
- 4
Routed to a specialist
Axis dispatches the request to the right specialist agent — a subject trainer, grader, lab supervisor, or certificate authority.
- 5
Specialist works
The specialist generates the answer, lesson, grade, lab guidance, or credential within its scoped capabilities and safety boundaries.
- 6
Output re-checked
The Safety Guardian and Axis review the specialist output before it reaches you, filtering unsafe content and adding defensive context.
- 7
Delivered with context
You receive the result, always paired with detection and defense framing so offensive concepts remain educational and lab-only.
- 8
Decision recorded
Axis logs the routing, rationale, confidence, and outcome so every automated decision is auditable and explainable.
- 9
Skills updated
Your outcomes feed your skills matrix and student memory, so the next request is personalized to your strengths and gaps.
- 10
You can ask "why?"
Any grade, recommendation, or restriction exposes the AI reasoning behind it and an automated appeal path.
The safety pipeline
Nothing reaches a specialist — or reaches you — without passing safety review twice.
Stage 1
Intent classification
Prism — Intent Classification Agent
Prism identifies the true request and flags ambiguous or dual-use asks for closer review.
Stage 2
Safety screening
Guardian — AI Safety & Ethics Guardian
The Guardian blocks real-world targeting, unauthorized access, and destructive requests before any specialist runs.
Stage 3
Safe-alternative redirect
Axis — AI Agent Orchestrator
Blocked requests are answered with an equivalent, legal exercise in a controlled lab instead of a refusal alone.
What the AIs do
Four core responsibilities, all performed by coordinated AI agents.
Teach
Specialist trainers deliver personalized lessons across ethical hacking, cloud, forensics, secure coding, and defense.
Grade
Graders assess quizzes, exams, and reports with multi-pass integrity checks and written, appealable feedback.
Supervise labs
A lab supervisor watches every command in isolated environments and blocks anything that leaves the training network.
Certify
A certificate authority issues verifiable credentials that clearly disclose AI-conducted instruction and assessment.