cluster

Deploy new applications to the Kubernetes cluster via FluxCD GitOps. Creates HelmReleases, Kustomizations, and namespace configs following repo conventions. user: "Deploy jellyfin" → HelmRelease with app-template, HTTPRoute, persistence user: "Add uptime-kuma" → Deploy with probes, route, resource limits user: "Install prometheus exporter" → HelmRelease with custom scrape config Use proactively when the user mentions deploying, installing, adding, or setting up an application.

Manage kopiur Kopia-based backups and restores for Kubernetes PVs in this cluster. user: "check backup status for app X" → describe the SnapshotPolicy/SnapshotSchedule user: "restore from backup" → follow the restore workflow below user: "trigger manual backup" → `kubectl kopiur snapshot now --policy <app> -n <ns> --wait` user: "backup failing" → delegate to debug-cluster for mover pod logs Use when the user mentions backups, restores, snapshots, SnapshotPolicy, SnapshotSchedule, Restore, ClusterRepository, PVC recovery, or disaster recovery. This cluster uses kopiur (home-operations, kopia-native) — not VolSync, not restic. Migrated 2026-07-12.

Diagnose and resolve Kubernetes cluster issues using structured fact gathering and 5-Whys root cause analysis. user: "pod CrashLoopBackOff" → describe pod, logs, events; GitOps fix in repo user: "HelmRelease not reconciling" → flux get/describe, reconcile with source user: "service unreachable" → endpoints, HTTPRoute, workload health user: "backup mover failing" → often delegate with backup-restore context Use for pod failures, OOM, Flux reconciliation failures, unreachable services, or node issues.

Research Kubernetes GitOps patterns from public GitHub repositories tagged `k8s-at-home` and adapt findings to this cluster. Use before adding or upgrading apps, writing HelmRelease/Kustomization/HTTPRoute/PVC/backup manifests, comparing chart values, or investigating how homelab clusters configure an application. user: "How do homelab clusters deploy X?" → Search topic repos, cite exemplar manifests user: "Find k8s-at-home examples for jellyfin" → Shortlist repos, compare HelmRelease patterns user: "Chart values for app-template like others use" → Code search scoped to top repos Prefer this before inventing Kubernetes YAML when the app may exist in the k8s-at-home ecosystem.

Use git worktrees for isolated, parallel agent work without polluting the main working tree. user: "work on feature X" → create worktree and branch under .worktrees/<task> user: "experiment with Y" → detached worktree for safe trials user: "parallel task" → separate worktree per concurrent task Triggers: worktree, isolated work, parallel agent, experimental branch, feature branch.

Review GitOps Kubernetes PRs or local diffs: YAML format, naming, HelmRelease patterns, SOPS/security, structure, and build validation. user: "Review this PR" → six parallel subagents, aggregate under .agents/pr-review/pr-<id>/ user: "Check my app config" → phase 3 (best practices) subagent user: "Are secrets encrypted?" → phase 4 (security) subagent user: "Validate before CI" → phase 6 (validation) subagent user: "Review my local changes" → PR_ID=local-changes, staged + unstaged diff Use proactively for K8s app/Flux/HelmRelease changes, infrastructure edits, or pre-commit diff review.

Summarize Kubernetes cluster health from Prometheus: firing alerts, CPU/memory hotspots, and workload signal. Default lookback is 30 minutes unless the user specifies another window. user: "cluster health" / "health snapshot" → alerts + top CPU pods over $WINDOW user: "firing alerts" → list_alert_rules or ALERTS PromQL fallback user: "top CPU pods" / "how's the load" → container_cpu usage queries (cores, not sec/min) user: "is karakeep healthy?" → filter namespace/pod + correlate alerts Prefer observability MCP (Grafana / ToolHive group): read session tool schemas before calling.