Player Intrusion: The player must be within trigger area.
Self Entry: The AI must be in the trigger area.
This determines the extent to which the actions override other things the AI might perform, e.g. patrolling, reacting to the player etc.
Default/None: Overrides patrolling, making it suitable for making a guard pause on patrol. Does not override alert reactions.
Very high: This overrides most actions (including fighting and fleeing), but not death.
Absolute: Nothing will stop the AI, not even death.
The other settings need to be examined
These two values (in feet) define a cylinder that the AI or the Player must be inside.
Not to be confused with alertness. The AI can be aware of the player without being alert if the two are on the same team, or the AI’s alertness is capped.
None: The AI does not have to be able to see the object, and physical obstructions (e.g. terrain) will not prevent the actions.
Line of sight: The object must be in the AI’s field of view, and there must be at least a small amount of light.
Raycast: This seems to be the same as Line of sight. Further testing required.
Specifies the alert levels at which the actions can be performed. An AI with the M-Front Gate Guard metaproperty can be alert without doing anything that might take priority over the actions.
This allows the AIWatchObj link to be deleted. This is useful if the AI is to only perform the actions once.
After Trigger: If the actions are triggered, the link will be deleted when the AI stops performing them, regardless of whether or not they were completed. In other words, the AI could get into a fight half way through, thus aborting the actions, and the link will be deleted. It will also be deleted if the actions are completed.
After Completion: Deletes the link if the actions are completed.
When the actions have been completed, this option removes any duplicate AIWatchObj links (i.e. same source and destination). If an AI has different responses to the same object for different triggering conditions, this option could be used to ensure that only one set of actions are performed.
By default, when the actions are being performed, the AI will keep checking the conditions, and will stop the actions if any one of them is no longer true, e.g. the AI becomes too alert. If this option is used, those checks will stop being made once the actions have started.
Once the actions have finished, this value (in milliseconds) determines how long the AI should wait before checking the conditions again. This allows an AI to escape from a TrolPausePt, for example. The default value of 10 seconds is plenty of time for an AI to resume patrolling at get out of the trigger area before checking the conditions again.
If the actions are aborted, this value (also in milliseconds) determines how long the AI should wait before checking the conditions again.
These are the actions to be performed. The correct name for them is a Pseudo-Script.