Skip to content

v0.0.1

Choose a tag to compare

View all tags
@droot droot released this 02 Apr 22:39
· 305 commits to main since this release
v0.0.1
ce79d89
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Changelog

  • 6fa9519 Able to eval strategies
  • a451069 Add basic project files
  • 224d8f1 Add simple kubernetes container for serving models
  • 7372f95 Add tools package (#43)
  • 5eb1e8c Add version support
  • b25da62 Added 'ask-for-confirmation' when resources are affected (#32)
  • fc17bd9 Added MCP server for kubernetes
  • c430dc2 Added agent that uses function calling
  • fa33d5f Added analyze subcommand to k8s-bench to generate eval report (#47)
  • e53b377 Added k8s-bench
  • ce79d89 Added release automation (#86)
  • e2ddea4 Added retry support to chat interface (#78)
  • 2eea8f1 Added support for direct stdin support (#12)
  • 0b0b9fa Basic configuration in ~/.config/kubectl-ai/config.yaml
  • 7f8b9b2 Better configuration loading (#70)
  • 08a94e5 Cleaned up the code (#76)
  • d5e1c44 Don't logspam on Control-D or stdin EOF
  • f6dafb4 Fix "typo" in eval (#66)
  • a564c36 Fix typo in prompt
  • d32b9e1 Fixed bug in retrying logic (#79)
  • 0d23ab9 Fixed misc bugs in mcp server code (#83)
  • 3a4ece7 Fixed verification for fix-service-routing eval task
  • 2291235 Have functions return typed output
  • fcee398 Made "query" a positional argument and updated README.md (#11)
  • a6c0f7a More structured logging (#34)
  • 90cb9e4 Nicer String() formatting in gollm (#20)
  • bebcd88 Pretty print tool invocation
  • 1b343a0 Refactor LLM interface into a separate package
  • 47ac6d5 Refactor terminal rendering into its own package
  • 4bdefbd Refactor tool calling for more logging
  • dfd403a Refactor: move strategies into their own packages
  • 07650c5 Refactor: start to split out strategy
  • 698400d Removed Strategies to have one agentic loop (#69)
  • f557b74 Schema support for bool and int types
  • e416f98 Set a default kubeconfig for k8s-bench
  • 83d60a6 Switched to official genai sdk (#72)
  • 284c5d9 Updated k8s-bench results (#74)
  • abe3fd5 Use a conversational logical model
  • b246fc9 Use gemini-2.5-pro as default gemini model (#82)
  • e649ae8 Use kubectl --wait in eval scripts
  • 2f1c1e9 Write a structured journal of what we do
  • a91e5b2 added a section on how to use the kubectl ai plugin invocatio method (#54)
  • 6794156 address json parsing nits (#9)
  • 6cfb265 bugfix: update the model during the interactive session
  • f705349 build(deps): bump golang.org/x/net from 0.35.0 to 0.36.0 (#44)
  • 9f2f69b build(deps): bump golang.org/x/net from 0.35.0 to 0.36.0 in /gollm (#53)
  • cd7315f build(deps): bump golang.org/x/net in /kubectl-utils (#60)
  • 642d14d chat mode: don't echo responses (#33)
  • 45ae5f7 chore: add license headers (#25)
  • 17676e6 chore: add simple presubmits (#24)
  • 233893e chore: fix compilation error
  • 02ddd5d ci: add tests for formatting and autogen (#26)
  • 6418a72 ci: fix name of main branch (#35)
  • babdff8 embed the prompt templates using go:embed and minor fixes
  • 94602fa eval: create helper script to run evals
  • 0846b7e evals: able to check output
  • 31ac32b evals: add eval to list images for pods
  • f2af6d9 evals: added create resources evals (pod and networkpolicy) (#46)
  • 090ebd0 fix: don't use kubectl run --timeout (#31)
  • 9db3554 gollm: add a simple scheme for env-var based LLM configuration
  • dec5070 gollm: remove duplicate methods (#23)
  • 34ca86d gollm: support for llama.cpp / llama-server (#45)
  • 08d69a5 gollm: support restricting output to a response schema (#21)
  • 1c9d27e improve klog usage
  • 61f48e1 initial import of kubernetes AI agent
  • 59759b2 k8s-bench: add go.mod replacement for root module
  • a53f173 k8s-bench: allow running each task in its own cluster, easily
  • fb2e51e k8s-bench: change syntax to support multiple-step interactions
  • 4cbd5f1 k8s-bench: start to support multi-step eval
  • d28a5c4 k8s-bench: start to write structured results
  • cc4325b k8sbench: log stdout/stderr to file (#30)
  • c0c1e2f k8sbench: switch to sigs.k8s.io/yaml
  • aa4c314 kubectl runner handles input command correctly
  • 8af501a modelserving: add cuda support
  • c1bc5e3 ollama support
  • a66d325 react prompt improvements for emitting json codeblock
  • 63048d7 support KUBECONFIG env variable
  • 9226408 tools: add simple tool to call trivy
  • 121754c use kubectl wait
  • 9efdeb6 use kubectl wait instead of timeout command (#36)
  • 7f502ae utils: create simple util to verify status with CEL

Released by GoReleaser.

Assets 11
Loading