A fuzzy-first interactive Kubernetes resource navigator
kf lets you fuzzy-search every resource across every namespace in your cluster from a single terminal window. Select one or many, then describe, exec, tail logs, delete, port-forward, restart, or dump YAML — all without typing a single kubectl command.
- Fuzzy search everything — pods, deployments, services, secrets, configmaps, nodes, namespaces, PVCs, jobs, cronjobs, statefulsets, daemonsets, ingresses — all at once
- Live preview pane — inline
describe, YAML manifest, or pod logs, cycled withctrl-p - Live watch — resources appear and update in real time as the cluster changes; deleted resources show
[DELETED] - Unhealthy-first ordering —
CrashLoopBackOff,Error,ImagePullBackOffpods surface to the top automatically - Color-coded status — red for critical, yellow for warning, green for healthy, dimmed for deleted
- Multi-select bulk actions —
tabto select multiple resources, then describe/delete/restart them all at once - Multi-cluster support — watch all kubeconfig contexts simultaneously with
--all-contexts, or switch contexts interactively withctrl-x - Context persistence — last-used context is remembered across sessions
- Demo mode — works without a cluster; shows sample data so you can explore the UI
A full user manual — installation, every keybinding, all actions, architecture, multi-cluster workflows, and the story of how skim shaped the design — is available as a compiled PDF:
📄 docs/manual/kuberift-manual.pdf
- Rust toolchain (
cargo) — to build from source kubectlin$PATH— used for all actions (describe, logs, exec, delete, etc.)- A valid kubeconfig (
~/.kube/configor$KUBECONFIG) — optional; demo mode activates automatically if absent
git clone https://github.com/syedazeez337/kuberift.git
cd kuberift
cargo build --release
# Binary is at target/release/kf
# Optionally move it somewhere on your PATH:
sudo mv target/release/kf /usr/local/bin/kfNote: The skim dependency is pulled automatically from a patched git fork during
cargo build. No separate clone is required.
kf [RESOURCE] [OPTIONS]
kfOpens the TUI with every resource type streaming from the current kubeconfig context.
kf pods # or: pod, po
kf deploy # or: deployment, deployments
kf svc # or: service, services
kf sts # or: statefulset, statefulsets
kf ds # or: daemonset, daemonsets
kf cm # or: configmap, configmaps
kf secret # or: secrets
kf ing # or: ingress, ingresses
kf node # or: nodes, no
kf ns # or: namespace, namespaces
kf pvc # or: persistentvolumeclaim
kf job # or: jobs
kf cj # or: cronjob, cronjobskf --context my-prod-clusterOverrides both the kubeconfig current context and the last-saved context.
kf --all-contextsStreams resources from every context in your kubeconfig in parallel. Each item is prefixed with its cluster name (color-coded per cluster).
| Key | Action |
|---|---|
| Type anything | Fuzzy filter the list in real time |
↑ / ↓ |
Move cursor |
tab |
Toggle selection on current item (multi-select) |
esc |
Quit |
| Key | Action | Multi-select |
|---|---|---|
enter |
kubectl describe |
✓ |
ctrl-l |
Stream pod logs (--tail=200) |
✓ |
ctrl-e |
kubectl exec -it into shell |
— |
ctrl-d |
Delete with y/N confirmation |
✓ |
ctrl-f |
Port-forward (prompts for local/remote port) | — |
ctrl-r |
kubectl rollout restart (deploy/sts/ds) |
✓ |
ctrl-y |
Print YAML to stdout | ✓ |
| Key | Action |
|---|---|
ctrl-p |
Cycle preview mode: describe → yaml → logs |
ctrl-x |
Open context picker — switch cluster without restarting |
The right-hand preview pane updates as you move the cursor. Press ctrl-p to cycle through three modes:
| Mode | Content |
|---|---|
describe |
kubectl describe <resource> output |
yaml |
kubectl get <resource> -o yaml |
logs |
Last 100 lines of pod logs (pods only) |
kf --all-contextsAll contexts from your kubeconfig are loaded in parallel. Items are prefixed with the cluster name:
pod prod-cluster/default/api-server-7d9f Running 2d
pod staging/default/api-server-5c2a Pending 5m
Each cluster gets a distinct color so items are immediately identifiable.
Press ctrl-x while kf is running to open a secondary fuzzy picker showing all your kubeconfig contexts. Selecting a context restarts the resource stream from that cluster. The selected context is saved to ~/.config/kuberift/last_context and restored on the next launch.
| Color | Meaning | Example statuses |
|---|---|---|
| Red | Critical — needs attention | CrashLoopBackOff, Error, ImagePullBackOff, OOMKilled, Failed, Evicted |
| Yellow | Warning — transitional | Pending, Terminating, Init:0/1, ContainerCreating |
| Green | Healthy | Running, Succeeded, Active, Bound, ClusterIP |
| Gray | Gone | [DELETED], Unknown |
Unhealthy resources (red) automatically sort to the top of the list so critical issues are visible immediately without scrolling.
If kubectl cannot connect to a cluster (no kubeconfig, invalid context, or network error), kf falls back to demo mode and displays 11 sample resources so you can explore the interface:
KUBECONFIG=/nonexistent kf
# [kuberift] No cluster (...). Showing demo data.kf -n production # restrict to the 'production' namespace
kf --read-only # disable delete, exec, port-forward, rollout-restart
kf --kubeconfig ~/alt.yaml --context staging # use an alternate kubeconfig| File | Purpose |
|---|---|
~/.config/kuberift/last_context |
Last-used context, restored on next launch |
$XDG_RUNTIME_DIR/<pid>/preview-mode |
Preview mode state (0=describe, 1=yaml, 2=logs) |
$XDG_RUNTIME_DIR/<pid>/preview-toggle |
Shell script installed at startup for ctrl-p |
MIT — same as skim, the underlying fuzzy engine.