Unfortunately, someone at Alfresco without comprehensive understanding of how node IDs can be structured in the backend implemented that regex-based extraction of node ID from the URL, which is incompatible with virtual node IDs. In a project I worked on together with an Enterprise partner during end of last year, we ran into the same issue, and I believe in the end we switched most of our navigation to being path based wherever smart folders might be involved.
I am not sure whether the ADF developers of the Enterprise partner did report this issue in a public ADF/ACA project - at least I could not find it just now. It is likely they addressed it via Support.
The pattern of a virtual node ID can be somewhat dynamic depending on how deep the navigation hierarchy is that is defined in the smart folder template + into which the user has navigated. Technically, Alfresco also supports multiple virtual node ID formats through its ID virtualisation mechanism, and potentially (though unsupported / undocumented, and requiring a bit of trickery) it could be extended by 3rd party implementations.
Also, even regular node IDs can literally be anything, and the default handling of ADF is broken even for some nodes generated when you install Alfresco as an unmodified product. Some nodes in the Data Dictionary structure have hard-coded IDs which are readable names instead of UUIDs.