That’s not generally the deployment model we use / recommend.
Generally speaking, one agent per queue - what you would likely , and inevitably see if they are running separate is both try to submit a job a fraction of a second apart (because they don’t talk to each other), and you’ll get a traceback and “WARNING’ that this flow runhas already been submitted.
The agents are very lightweight, and there should generally not be a need or requirement to duplicate agents listening to the same work-queue, unless you have a good use case we can better understand?