feat(plugins): add plugin kubectl-ai, Integration for Intelligent Operations

[oopslink]

Overview

This PR integrates kubectl-ai functionality into k9s, bringing LLM-powered intelligent Kubernetes operations to the terminal UI experience.

What is kubectl-ai?

kubectl-ai is an AI-powered Kubernetes assistant that leverages Large Language Models (LLMs) to translate natural language requests into kubectl commands and execute Kubernetes operations 1116. It acts as an intelligent interface between users and Kubernetes clusters, eliminating the need to manually craft complex kubectl commands.

Features Added

• Natural Language Interface: Users can describe operations in plain English within k9s
• Smart Command Translation: AI automatically converts natural language to appropriate kubectl commands.

Example Usage

  Hey there, what can I help you with today?


>>> list all finished job in ops namespace
  Running: kubectl get jobs -n ops --field-selector status.successful=1 -o jsonpath='{range .items[?(@.status.succeeded==1)]}{.metadata.name}\n{end}'


  Here is the list of all finished (successful) jobs in the "ops" namespace:

  • auto-pause-29189260
  • cleanup-29189260  
  ...
  • superset-init-db

    Let me know if you need any more details! 😊


>>> cleanup those jobs which name start with 'cleanup'
  Running: kubectl delete jobs -n ops -l job-name-prefix=cleanup --field-selector metadata.name=cleanup

  Do you want to proceed ?
  1) Yes
  2) Yes, and don't ask me again
  3) No
  Enter your choice (1,2,3): 2

  The command to delete jobs with names starting with "cleanup" using the
  label selector did not find any resources because those jobs don't have such
  a label. I will delete the jobs based on name prefix using a different
  approach by listing and deleting them explicitly. Proceeding with this now.


  Running: kubectl get jobs -n ops -o jsonpath='{.items[?(@.metadata.name.startsWith("cleanup"))].metadata.name}' | xargs -r -n 1 kubectl delete job -n ops


  The JSONPath expression I used to filter jobs by prefix is invalid. I will
  try another approach: first list the jobs with name starting with "cleanup"
  in the "ops" namespace, then delete them one by one. I will do this in two
  steps.


  Running: kubectl get jobs -n ops -o jsonpath='{range .items[?(@.metadata.name =~ /^cleanup/)]}{.metadata.name}\n{end}'


  The jsonpath filter for regex match is not supported by kubectl. I will use
  a different approach: get all jobs in the ops namespace, then filter the
  names starting with "cleanup" using shell commands, and delete those jobs.
  Proceeding with this approach now.


  Running: kubectl get jobs -n ops -o jsonpath='{.items[*].metadata.name}' | tr ' ' '\n' | grep '^cleanup' | xargs -r -n 1 kubectl delete job -n ops


  All jobs in the "ops" namespace with names starting with "cleanup" have been
  deleted successfully. If you need any further assistance, feel free to ask!
  😊

>>> list jobs in ops namespace
  Running: kubectl get jobs -n ops


  Here is the list of jobs in the "ops" namespace:

  • auto-pause-29189260
  ...
  • superset-init-db

  If you want me to filter or show specific details, just let me know! 😊

>>>

Security & Production Readiness

⚠️ Important Notice:

The intelligent capabilities and security depend on the LLM model used by the kubectl-ai project. As this technology is still maturing, please DO NOT USE in production environments.

💰 Cost Considerations:

• Token Usage: The system may examine pod logs, events, and cluster state during analysis
• Potential High Costs: Large clusters with verbose logging can consume significant LLM tokens
• Recommendations:
  • Monitor your API usage and billing
  • Set up cost alerts and limits
  • Consider using log filtering or sampling for large deployments
  • Test with small clusters first to estimate costs

Dependencies

Requires kubectl-ai to be installed and configured

Breaking Changes

None. This is an additive feature that doesn't modify existing k9s functionality.

[derailed]

@AaronLinOops Nice! Looks like we have conflicts ;( Can you take a peek?

[oopslink]

@AaronLinOops Nice! Looks like we have conflicts ;( Can you take a peek?

@derailed My apologies for the delay! ^^; the conflict is now resolved! :-)

[oopslink]

It looks like dockercloud was unavailable at the time causing the workflow to fail.

{"error": "Our service is temporarily unavailable. We'll be back soon!"}

[derailed]

@AaronLinOops Tx for the updates!

[derailed]

For those that don't read the main readme ;(
We should have a brief header with the project url and how to install it

[github-actions]

This PR is stale because it has been open for 30 days with no activity.