What we capture

Every look, action and decision inside a simulation is recorded, not just the final score.

A learner's session produces a detailed record of what they did, in what order, and how long each step took. That record is available in the Insights dashboard and can be exported for your own analysis.

This page describes what is captured in plain terms. For field names and formats, see How logging works, the Event Reference and the Attribute Reference.

Attention and gaze

Where the headset was pointed throughout the run. Head position and orientation are sampled continuously, alongside both controllers, which lets a session be replayed and shows which parts of the environment a learner spent time facing.

Object interaction

Every item picked up, put down, snapped into place, removed, used and released. Collisions between objects are captured too.

Each interaction names the objects involved, so you can see that a learner grabbed the wrong tool before finding the right one.

Movement and navigation

Where a learner moved, where they stopped, and how long they stayed. Teleports and continuous position are both recorded, so the path through the space can be reconstructed.

Menu presses, button clicks and keyboard input are captured on the platforms that use them.

Decisions and progression

Simulations are built as state machines, and every state entered, exited and transitioned is recorded, along with the condition or trigger that caused it.

That means the record shows not just that a learner advanced, but what they did to advance, and which conditions they satisfied or failed on the way.

Skipped steps, repeated attempts and hesitation are all visible in the sequence.

Speech and dialogue

In conversational simulations, the full transcript is recorded. Each learner utterance is stored verbatim, along with how the system interpreted it, the response given back, and how long that response took to arrive.

Where a simulation uses an AI character, the character's configuration and the conversation history are recorded with the exchange.

Assessment

Quiz questions record the question asked, the choices offered, the answer selected and whether it was correct.

Skills and mastery

Simulations can be mapped to a skill ontology. As a learner acts, observed skill values are updated and propagated up that ontology, and each update is recorded with its previous and new value.

This is what produces mastery over time rather than a single score at the end.

Feedback and scoring

Feedback shown to the learner is recorded, including the messages given, the timers behind them and how long the run took.

Session integrity

Alongside learner behaviour, each session records the technical conditions it ran under: device and platform, framerate, how long assets and scenes took to load, and any errors that occurred.

Where this applies

Capture is consistent across the platforms HyperSkill runs on, including VR headsets, desktop, web, mobile and mixed reality. Some categories depend on the hardware. Controller interaction and room scale movement apply where the device supports them.