Back in the 1970s I wrote something called BATSTAT which accomplished a somewhat similar task.
The goal was to be able to know which short batch jobs would result in printed output in the next half hour or so.
The key was to use a detached VM/CMS virtual machine that submitted a batch job primarily consisting of a “/*$” command (e.g. /*$DJ1-9999).
The output would be routed to the virtual machine that was awaiting the response (in its Virtual Reader).
The Virtual Machine would read in the response and reformat it, then send it to the printer. The “distinctive looking” printout would be posted near the place where printouts were filed alphabetically.
Does this give you a starting point ?