Skip to content

Lab Types

Every lab in Rogue Arena has an environment type, shown as a labeled badge on the scenario. The type doesn’t affect the infrastructure: the same machines and network are deployed whatever the type. It controls what you can see and do from the start, meaning whether the target machines are visible, and whether tasks are open from the beginning or revealed as you progress. The scenario’s author sets the type during Lab Creation.

TypeBest forVictim VM visibilityTasks
Mini LabOpen exploration, tool testing, demosAll VMs visibleAll unlocked from the start
ScenarioGuided, discovery-driven challengesLocked until you complete the scenarioLocked; unlock as you progress
CourseStructured training with course contentLocked until you complete the scenarioLab tasks locked; other course content usually visible
Certification TestAssessmentsLocked until you complete the scenarioUnlock by the test’s own criteria
Model TestingTesting AI agentsLocked until you complete the scenarioEval hops score the agent’s progress

In a Mini Lab, everything is open. All VMs are visible, including the target (victim) machines, and every task is unlocked from the start. Nothing is gated behind progress. This is the sandbox mode: the whole environment is available as soon as it comes up. Use it for free-form exploration, testing tools, quick demos, or practicing against a network you already understand.

A Scenario is a guided challenge. Victim VM visibility is locked until you complete the scenario, and tasks are typically locked, unlocking one step at a time as you progress. You earn visibility and next steps by doing the work, so you discover the lab as you go. Scenarios suit red-team-style challenges and progressive labs where the reveal is part of the point.

A Course locks victim VM visibility and its lab tasks, but usually leaves the rest of the course content visible from the start.

A Certification Test is an assessment with its own unlock criteria. Its badge reads Certification, only Rogue Labs staff can set this type, and snapshots aren’t available in it.

A Model Testing environment is built for Model Testing: an AI agent works the range while eval hops score how far it gets.