Aadhav Vignesh

> For MLFLow we may use this https://www.mlflow.org/docs/latest/python_api/mlflow.tracking.html#mlflow.tracking.MlflowClient.download_artifacts but I wonder what to use with Kubeflow maybe a similar approach to [kubeflow/pipelines#3814 (comment)](https://github.com/kubeflow/pipelines/issues/3814#issuecomment-633778570) ? Looks good to me; but the...

I'm working on it, I found some flaws: - The PAT is **always read through user input**; there is no option of passing PAT through a flag. - I tried...

If you want me to add this feature, could I open up a PR today?

**Update:** I wrote a minimal example of setting up a `k3ai` cluster using Github Actions. You can see it action [here](https://github.com/burntcarrot/k3ai-actions-test/runs/4126233537?check_suite_focus=true). 🥳 *The current workflow doesn't use the binaries, as...

Yes, I know about the "extracting the config" part. This was just a proof of concept; I was excited to try it on Github Actions! 😋

Keeping this open as we still have to deal with a complete end-to-end workflow. (Artifact collection support is missing)

I use `kind` for local cluster development, but I'd like more priority for `k3d` than `k3s`. `k3s` doesn't work well with WSL2.

Opened #25 in favor of this issue.

On a side note, can we add the Dockerfile for `k3ai-executor` to the repo? Code execution relies on the executor image, so it makes sense to make it available.

I don't think auto reports through PRs sounds good. Opening an issue also looks messy, but [this action](https://github.com/peter-evans/create-issue-from-file) allows automatic labeling of issues related to automated reports. Since we have...