Error: starting k8s agent: initializing chat session: Initialize not yet implemented for ollama

[infinitydon]

Hi,

I am currently trying to use kubectl-ai with ollama running on a remote system and I am getting the following error:

export OLLAMA_HOST=http://192.168.1.155:11434/ kubectl-ai --llm-provider ollama --model gemma3:12b-it-qat

Error: starting k8s agent: initializing chat session: Initialize not yet implemented for ollama

I am able to conenct to the ollama endpoint:

curl http://192.168.1.155:11434/
Ollama is running

kubectl-ai version is:

kubectl-ai version
version: 0.0.22
commit: f1516148e4bd9a39f0c3af63256fcf8c6e521ea9
date: 2025-08-14T18:00:57Z

Not sure if this is a bug?

[droot]

Aaah.. this is a bug that got introduced recently. If you are blocked on it, pl use v0.0.20

@noahlwest

For llm providers that doesn't support have session persistence in:

• If user opts in to use session, then we should print warning.
• for in-memory, there shouldn't be any errors at all.

[noahlwest]

Ack, this one is from one of my recent changes. The issue here is is an oversight from adding the in memory chat store: calling Initialize no matter what, which would've been caught by a nil check when there was no in-memory store. Made #492 to log the error and return nil instead as a quick fix.

[droot]

Thanks @noahlwest for the quick turnaround. https://github.com/GoogleCloudPlatform/kubectl-ai/pull/492 is now merged.

@infinitydon it will be available in the next release (v0.0.23).