Thank you for adding support for modifying existing manifests. I would like to see an extension of this support to enable support for mulitple files, and directory awareness. The specific use case is to take an existing manifest set with eg deployment, ingress and service manifests and using a prompt to “upgrade this” to create a kustomize compatible output fileset / structure using bases and overlays for production, staging etc. Doing this requires, at a minimum, geneation of output into a kustomize compatible folder structure. This could be done using a compressed format such as zip / tar as input and output options or it could be done by modifying kubectl-ai to be multi-file aware in adition to input / output stream capable. Supporting multiple input files in a directory structure would also be nice to be able to modify the generated output where there is a directory structure.
I might be willing to implement this if I could get some design help to figure out the right approach.
Are you willing to submit PRs to contribute to this feature request?
What kind of request is this?
New feature
What is your request or suggestion?
Thank you for adding support for modifying existing manifests. I would like to see an extension of this support to enable support for mulitple files, and directory awareness. The specific use case is to take an existing manifest set with eg deployment, ingress and service manifests and using a prompt to “upgrade this” to create a
kustomizecompatible output fileset / structure using bases and overlays for production, staging etc. Doing this requires, at a minimum, geneation of output into a kustomize compatible folder structure. This could be done using a compressed format such as zip / tar as input and output options or it could be done by modifying kubectl-ai to be multi-file aware in adition to input / output stream capable. Supporting multiple input files in a directory structure would also be nice to be able to modify the generated output where there is a directory structure.I might be willing to implement this if I could get some design help to figure out the right approach.
Are you willing to submit PRs to contribute to this feature request?