cfgmgr:

• Merge/Delete kubeconfig file to ~/.kube/config

secrets:

• Base64 decode secrets data

entry:

• kubectl exec bash/sh in container
• kubectl entry <fzf query term> [-n namespace] [-c container]

log:

• kubectl logs with fzf
• kubectl log <fzf query term> [-f] [-p] [-n namespace] [-c container] ...

desc:

• kubectl describe with fzf
• kubectl desc [resource] [fzf query term]

g:

• kubectl get with fzf
• kubectl desc [resource] [fzf query term] [options]

diff-rs:

• kubectl diff-rs
• kubectl diff-rs LAST LAST-3

rs:

• A Deployment's whole rollout history: ReplicaSets oldest->newest, then a diff of every CONSECUTIVE revision (each block = what that rollout changed). diff-rs compares TWO; this walks all retained revisions.
• kubectl rs <deployment> [-n NS | -A] (single token — kubectl-rs; kubectl reads each dash as a word boundary, so a kubectl-rs-timeline name would dispatch as kubectl rs timeline)
• filters by ownerReferences (correct lineage, no cross-deployment interleaving); sanitizes via kubectl neat (--no-neat for jq scrub, --raw for full yaml)
• diffs render with dyff (semantic, path-addressed) when installed, else colorized diff -u; force the latter with --plain
• auto-pages on a TTY via $PAGER (else less -RF, else bat); piped/redirected output is never paged; -P/--no-pager to disable
• --no-diff: timeline table only
• unknown flags / anything after -- pass through to kubectl get rs (--context, --kubeconfig, -l ...)

recreate:
kubectl recreate -f

cordonx:

• kubectl cordonx [node2 ...]
• kubectl cordonx --undo

find:

• kubectl find [pattern2] [...]
• kubectl find po -i
• kubectl find po -s

vatt:

• pvc to volumeattachment
• kubectl vatt <pvc_name_pattern>

prune:

• Force-delete pods whose NAME matches a pattern, with confirmation
• kubectl prune <pod-name-pattern> [-n NAMESPACE | -A]
• kubectl prune --all [-n NAMESPACE | -A] — delete all UNHEALTHY pods (READY not X/X or STATUS not Running/Completed/Succeeded)
• default: current namespace; always --force --grace-period 0

fdrain:

• Drain node(s), force-deleting pods the eviction API can't move because a PodDisruptionBudget would be violated (drain --force does NOT bypass PDBs)
• Only acts on ALREADY-CORDONED (SchedulingDisabled) nodes — cordon first; uncordoned matches are skipped. Prompts once for the destructive op
• kubectl fdrain <node> [node ...] — runs kubectl drain --delete-emptydir-data --ignore-daemonsets (defaults) per node in parallel, then force-deletes (--force --grace-period 0) any pod still pinned by a zero-allowance PDB
• kubectl fdrain -l <selector> — target every cordoned node matching a label selector (e.g. -l node.info/kubeletVersion=v1.28)
• kubectl fdrain -y ... — skip the destructive-op and per-pod confirmations
• unrecognized flags (e.g. --force) pass through to kubectl drain; options: --poll, --grace (let eviction run first), --timeout

force-sync:

• Annotate an ExternalSecret with force-sync= to trigger an immediate sync
• kubectl force-sync [-n NS] [query]
• picks the ExternalSecret via fzf (fuzzy match on the optional query)

grep:

• kubectl get <resource> | grep <query> shortcut; query matches any column
• kubectl grep [/] [query]
• kubectl grep pod 7d (any column: NAME, STATUS, AGE, ...)
• kubectl grep 'pod*' nginx (multi-resource via resource glob, parallel)
• kubectl grep '*/nginx' (any resource, row contains 'nginx')
• -c / --compact: <ns>\t<r>/<name> per line (awk/xargs friendly)
• -nr / --not-ready: drop healthy rows (Completed|Succeeded|X/X)
• -E / --regex: treat query as raw regex (anchors, quantifiers)
• unknown flags pass through to kubectl get (-owide, --show-labels, ...)