Hi Martin.
Yes, that with the port objects was quite surprising.
I had initially made an “Ollama only” package, but ran into this problem:
knime.extension.nodes.InvalidParametersError:
WARN LLM Prompter
The provided input is unknown and therefore incompatible with the expected type LLM
(got org.knime.python.ollama.models.ollama._chat.OllamaChatModelPortObject).
Connect one of the following nodes: HF Hub LLM Connector, HF Hub Chat Model Connector, HF TGI LLM Connector,
HF TGI Chat Model Connector, OpenAI LLM Connector, ... and 9 more.
But after moving the code into the original package, it worked together.
Thanks to the Forum for pointing this out!
Re-use Custom Port Object from existing extensions Python-based extensions - Node Development - KNIME Community Forum Cool person 
As far as I understand, KNIME must somehow extract which nodes implement the LLMPortObject/LLMPortObjectSpec in src.models.base from the package.
And since this port specification is not part of core KNIME, all objects has to refer to the same specification.
I KNIME haven’t made their code publicly available in a repository, then there really is only one way.
Ref: 4 steps for your Python Team to develop KNIME nodes | KNIME
External sharing to the whole KNIME community on KNIME Hub
requires sharing your code in a publicly available git repository
(e.g. BitBucket, GitLab, GitHub), providing some testflows, and
sending the link to KNIME community contributions mail.
We will run some automated tests and get back to you.
So
1: Fork the package and develop a community version.
2: Ask KNIME to add this as Community Extensions
It’s just a silly message…
Hey, don't use KNIME's Package, but
this community package instead.
It has more features and accepts PR from the community.