Skip to content

Roadmap

Upcoming (v0.10.x maintenance)

Staying on the v0.10.x line for a while to shake out bugs and collect real-world feedback from the helm-guard + TUI import/merge work that shipped in v0.10.0. No new headline features are planned during this window; patch releases will focus on:

  • Bug fixes surfaced from manual and production use.
  • Follow-up polish from .sisyphus/drafts/post-v0.10.0-review.md items not yet addressed.
  • CI / dependency hygiene (action bumps, goreleaser minor updates).

Larger planned features resume in v0.11.x — see Long-term below for the queue.

Long-term

Versioned documentation site via mike

The Pages site currently publishes the docs from the latest release tag (single version). Long-term we want a version switcher (dropdown showing v0.9.0, v0.9.1, v0.10.0, …, dev) via mike.

Plan:

  1. Add mike to requirements-docs.txt alongside mkdocs-material.
  2. Add extra.version.provider: mike to mkdocs.yml — the Material theme renders the switcher automatically.
  3. Extend .github/workflows/docs.yml:
  4. On push to master → mike deploy --push --update-aliases <date> dev.
  5. On tag push v* → mike deploy --push --update-aliases <tag> latest.
  6. mike keeps all versions on the gh-pages branch; switch the Pages source from the GitHub Actions artifact to the gh-pages branch.

Proposed next feature target once the v0.10.x maintenance window closes.

Cloud sync — pluggable state backends

State is already sync-friendly: stable-topology keys, updated_at timestamps, versioned schema. The Store interface in internal/state accepts alternative implementations without changes to the CLI or TUI.

Planned backends, in order of complexity:

  1. Git-backed — point at a private repo, commit on every write. Zero server infra. Ships first.
  2. S3-compatible — bucket + last-writer-wins via ETags. Works with Minio, R2, etc.
  3. Vault KV — for teams that already have Vault.

No credentials ever leave the local kubeconfig files — sync is metadata only.

Group-scoped alerts by tag

Instead of enabling alerts per-kubeconfig, apply them to every kubeconfig carrying a given tag (e.g. all prod-tagged). Add a tag_policies: section in the state file keyed by tag. Bigger design lift than the guard / sync items; most useful once cloud-sync lands and teams can share tag policies.

Shipped

Items that were on this roadmap and have since landed, for history:

  • v0.10.0 — helm values-path guard (two scopes, multi-pattern list, path-token global fallback, default ON via tri-state); TUI parity for import + merge; kcm doctor, kcm audit, kcm init wizard; --dry-run on every mutating command; stable-hash state keys made default; major refactor (state / cli / tui split into per-feature files); multi-arch Docker images published to ghcr with cosign keyless signing.
  • v0.9.2 — fish shell support; kcm context rename / kcm context delete with TUI keybindings (R/D/S); kcm starship prompt integration; testscript end-to-end suite; testable guard.Exec via stub kubectl; alert-indicator cleanup in TUI detail view; runtime/debug build-info fallback for go install; docs site now deploys from the latest release tag.
  • v0.9.1 — dynamic shell completion; kcm prune for stale/orphaned state entries; CLI golden tests with xdg.Reload isolation; Node 24-capable GitHub Actions.
  • v0.9.0 — initial public release: TUI + CLI for kubeconfig files, tags, destructive-action guard, shell integration (bash/zsh/pwsh), import/split/merge, Apache 2.0 license, multi-platform release workflow.