I’m not sure what you’re gaining by combining actions together that don’t share verbs (and therefore can’t be ambiguous to the parser)?
As a separate issue, it’s probably not good style to introduce constraints to your grammar tokens (such as [visible thing] instead of [thing]) unless there really is no other way of achieving what you want.
General rule of thumb is to keep things as broad as possible at parser level so that any vaguely sensible player input is accepted by the parser, then sort out details in action-processing where you have complete and transparent control of logic and responses.
The parser not matching [visible thing] gets you the default blank ‘You can’t see any such thing’ response, whereas ‘Before navigating something invisible:’ allows you to take full control.
As a further aside, the differences between [thing] (a thing in scope for the actor) and [visible thing] (a thing visible to the player) are subtle, unless the actor is not the player… (see here)
EDIT PS The differences between [visible thing] as a grammar token (as in 'Understand “destroy [visible thing]” as destroying) and ‘visible thing’ in an action definition (as in ‘Destroying is an action applying to one visible thing’) are not so subtle. The former means ‘a thing the player can see’ (thing here having the sense of ‘an instance of the thing kind, or of one of its sub-kinds, e.g. people or women or doors or containers)’, the latter means ‘any object in the universe, including rooms, regions, directions and objects out-of-play, and (confusingly) whether the player or anyone else can see them or not’.