Using multiple chat services in one Kernel for different skills

[VladTolm]

Maybe I'm doing something wrong, but it doesn't seem to work in the Python version reading the "default_backends" key for skills.

No matter what values I put there, the default chat service is still used.

In the tests and documentation I did not find examples of using different services for different skills

[aeharriman]

I am also encountering this issue. Using the kernel.set_default_chat_service() before calling semantic functions does not work either.

[aeharriman]

If anyone else ever runs into this, setting the default chat service actually ended up working. In my previous comment I was:

1. Using kernel.set_default_chat_service("service_id) when I should have been using kernel.set_default_text_completion_service("service_id")
2. Using it right before using kernel.run_async() with my semantic function I was trying to specify the service_id for.

What ended up working was using kernel.set_default_text_completion_service("service_id") before importing the respective skills into the kernel, like this:

kernel.set_default_text_completion_service("service_1")
kernel.import_semantic_skill_from_directory(skills_directory, "Plugin I wanted to use with service 1")

kernel.set_default_text_completion_service("service_2")
kernel.import_semantic_skill_from_directory(skills_directory, "Plugin I wanted to use with service 2")

[punitarani]

I still think it would be nice to have a param or config option for each skill to indicate the service preference