NX Command Prediction is consuming tokens when not in use.
The current behavior of consuming tokens on session is as designed for command prediction. This is because command prediction is required to be active such that:
* It can train on user command usage when it happens.
* Whenever a command is executed, it will create predictions of possible next commands based on past training.
* It will display these predictions in the command prediction bar such that the commands can be executed through the bar.
Because this behavior is automatic and ever present, the token must be consumed for the entirety of the session.
A user can disable Command Prediction through use of customer defaults, they appear aware of this possibility.As a reminder on how to disable Command Prediction through customer defaults, one must uncheck the "Predict Commands" checkbox under "Gateway > User Interface > Predict Commands" in the customer defaults menu, afterwards they will need to restart NX to get the final results. This fully disables command prediction from training on user command usage as well as disables the predicting and display of possible next commands.
Notes