Hi @podilam
I started working on YDK in 2017 and in my development did not hear about "dynamic API generation". It is possible that in early development there were some plans to do thins feature, but it has never got to actual implementation. And the reason is that following YDK releases (starting from 0.6.0) used libyang library to build internal view of the model from YANG models directly. The data models then built using Path API, which does not require pregenerated model API. I suggest you to read this YDK documentation to get familiar with Path API, and browse the unit tests for examples.
Although Path API method does not require pregenerated model API, it is still require knowledge of the YANG model that you are about to use in your application. Luckily some models like IETF, OpenConfig and Tail-f are very stable and actually preferred to be used by external users. Hence the programmatic way of monitoring and controlling the devices getting momentum.
The Yang Suit seems very convenient tool to build and execute RPCs with use of Netconf protocol. This could be helpful for issuing single RPCs, but it does not have programmatic approach that YDK is offers. Please correct me, if I am wrong. The YDK can be easily used in the way the YangSuite is working. You just need to have user defined XML payload and then build and issue RPC using YDK Executor Service. Examples of this approach you can find here.
I also wanted to mention Cisco developed ATOM tool, which dynamically learns the generated API and builds RPCs for testing XR devices. That is unique tool, which programmatically tests the devices without changing the core program, something that you are looking for.
Yan Gorelik
YDK Solutions