Trying to make things easier

I have several devices which ZigBee2MQTT recognises but the ZigBee binding doesn't seem to which is halting my move over to the binding.
In ZigBee2MQTT there is a procedure to create and add your own 'converter'. The way they do it is to modify the yaml. However, in reality, the converter is very similar to how we add channels to Things in the MQTT binding. As long as the scan can see the device we can then add our own channels to the device (rather than having to rely on a central repository) - the interview generally would inform users what endpoints were available and it's just a matter of making sense of them. As users add and define the devices they could then upload the code from the channels building a resource which can then be combined into the releases.
It might reduce the requests to support particular devices too.
It's just an idea...

Review the ZigBee binding's current device scan and interview flow, then compare how channels are added in the MQTT binding. Define how users would supply channels or converters for recognized devices and how those definitions could be contributed to releases; the issue is complete when this design is agreed and implemented.