anton

Kubernetes-layer health triage for Anton. Use when Codex checks whether the cluster is OK, Flux is healthy, platform controllers are ready, CNI or DNS is failing, cert-manager or ESO is stuck, gateways are broken, cloudflared is disconnected, or apps are unhealthy.

Intake gate for adding new system or infrastructure components to Anton. Use when deciding whether to adopt, install, evaluate, or learn a new app, operator, Helm chart, or platform component before scaffolding manifests. Applies concrete-need and honest-learning rubrics, checks containment and removal history, returns add/defer/reject with ADR-ready output, and stays read-only.

Deploy or onboard an owned Anton app through Flux image automation. Use for deploy, redeploy, force reconcile, deploy homepage, onboard app, new owned app, ADR 0024 image automation, ImageRepository loops, rollout watching, and public URL verification. Reports stuck reconciles and hands them to debug-flux-reconciliation.

Expose a workload for Anton access. Use when choosing between envoy-internal, Tailscale Ingress, Tailscale Service annotation, or explicitly approved envoy-external plus Cloudflare tunnel. Handles HTTPRoute authoring, DNSEndpoint for secondary domains, per-domain cert wiring, and route verification.

Operate Anton's time-boxed ClickStack logs and traces experiment. Use when handling HyperDX UI access, ClickHouse, Keeper, MongoDB, OTel collector health, Flux reconciliation, OTLP test signals, Lucene or SQL queries, ESO and 1Password credential rotation, Renovate chart or operator bumps, the 2026-08-02 ADR 0028 review, or ClickStack teardown. Keywords: clickstack, hyperdx, clickhouse, keeper, mongodb, otel, otlp, logs, traces, wide events, temporal traces, review-by, ADR 0028.

Operate the time-boxed ClickStack logs/traces experiment on anton (ClickHouse + Keeper + MongoDB + OTel collector + HyperDX, per ADR 0028). Use to access the HyperDX UI, send a test signal (pod logs via the bundled OTel collector, point Temporal's OTLP traces at it), run Lucene/SQL queries, check health of any ClickStack component, reconcile/debug the Flux flow, rotate the 1Password credentials via ESO, triage Renovate bumps for the two operators + chart, run the 2026-08-02 review-by checklist, or execute the exit/teardown runbook. Keywords — clickstack, hyperdx, clickhouse, keeper, mongodb, otel collector, otlp, logs, traces, wide events, lucene, temporal traces, teardown, exit plan, review-by, ADR 0028.

Scaffold a new Flux app for Anton. Use to add an app, deploy app, new helm chart, new namespace, scaffold flux app, or create external secret. Generates ks.yaml with postBuild, HelmRelease, OCIRepository, and optional ExternalSecret.

Add or replace a Talos node in Anton. Use to add a node, replace node, scaffold a new control plane, recover a dead node, plan node replacement like k8s-4, or edit nodes.yaml / talconfig.yaml. Generates patches, applies via talosctl.

Anton ADR lifecycle — author new architectural decision records, list existing ones by status or affects-category, and mark old decisions superseded. Use when capturing a decision (especially after `cluster-intake-gatekeeper` returns an ADD/DEFER/REJECT verdict), when reviewing prior decisions before changing direction, when checking if a candidate component has been removed before, or when promoting a decision out of memory into a durable record. ADRs live in `context/adrs/` and are immutable — supersession is the only way to change a decision. The ADR index is built by scanning ADR files directly and injected into every Codex session by `.Codex/hooks/inject_adr_index.py`. Keywords — ADR, architecture decision record, decision log, supersede, decision history, why did we, prior decision, recorded decision, MADR, immutable, intake handoff, cluster-intake-gatekeeper handoff, removal graveyard, reverted decision.

Upgrade triage for Anton — walks open Renovate PRs, flags supersessions, proposes a tiered merge order (patches → minors → CI majors → cluster majors), reads the Renovate Dependency Dashboard, and runs live-cluster deprecation audits with kubent/pluto/nova. Read-only. Use when asking "what can we upgrade", "check renovate PRs", "upgrade audit", "is anything outdated", "audit chart versions", "k8s deprecation scan", "what should we merge next", "are there superseded PRs". Hands Talos or Kubernetes version pin bumps off to upgrade-talos-or-k8s. Keywords — renovate, upgrade, audit, outdated, dependency dashboard, helm chart, kubent, pluto, nova, deprecation, chart drift, PR triage, merge order, supersede, bump, renovate bot

Configure Longhorn backup target (S3 or NFS), author RecurringJobs for snapshot/backup/filesystem-trim, and restore a volume from a backup. Use when wiring up backups for the first time, editing recurring schedules, or performing a DR restore. Not for PVC create (longhorn-volume-ops) or disk work (longhorn-node-ops).

Add or remove Longhorn disks on a Talos node, drain/evict replicas before maintenance, fix disk-routing mismatches, and integrate a replaced node into the Longhorn topology. Use for disk swaps, pre-reboot eviction, or when a node's longhorn disks don't match the Talos annotation. Pairs with add-or-replace-node for the Talos side.

Create, resize, clone, and tune per-volume Longhorn PVCs in anton. Use when adding a PVC backed by longhorn, overriding replica count or data locality for a specific workload, resizing an existing Longhorn volume, or cloning a volume from an existing PVC. Does not cover disk/node ops (longhorn-node-ops) or backup/restore (longhorn-backup-dr).

Triage ntfy.sh-routed alerts in Anton — identify which alert fired, why it fired (or why it didn't deliver), and propose a fix. Use when "got an ntfy alert", "alert just fired", "ntfy not delivering", "AlertmanagerClusterFailedToSendAlerts", "AlertmanagerFailedToSendAlerts", "code 40014", "attachments not allowed", "iOS push missing", "test the ntfy receiver", "send a test alert", "what just paged me", "is ntfy working". Combines kube-prometheus-stack Alertmanager API, the self-hosted ntfy server (ADR 0026), and the ntfy CLI for poll/publish probes. Read-only by default; proposes edits the operator applies.

Integrate an anton workload with kube-prometheus-stack. Use to add a ServiceMonitor, add a PodMonitor, expose metrics to Prometheus, author a Grafana dashboard (sidecar ConfigMap), write a PrometheusRule (alerts or recording rules), verify scrape targets, or debug why a metric or series is missing. Covers the observability stack installed per ADR 0007 / plan 0002.

Anton planner skill — author, update, and close multi-session initiatives (migrations, rollouts, long-running refactors) in `context/plans/`. Use when starting a multi-session initiative, tracking next steps on in-flight work, migrating a memory entry to a durable plan, closing a completed initiative, or reviewing what's open. Plans live at `context/plans/NNNN-kebab-slug.md` and are mutable — they capture execution state (what's next, what's blocked, log of decisions made during work) while ADRs capture immutable decisions (why). The active-plan index is built by scanning plan files directly and injected into every Codex session by `.Codex/hooks/inject_plans_index.py`. Keywords — plan, planner, initiative, track work, multi-session, next steps, checklist, migration plan, rollout plan, roadmap, in-flight, blocker, close plan, review-by, exit plan, timebox, memory-to-plan handoff.

Credential rotation for Anton — rotate age key plus SOPS key rotation across the repo, generate a new deploy key for Flux, rotate the 1password token for ESO, or rotate the cloudflare token for the tunnel. Preconditions, verification, rollback.

Rolling upgrade for Anton's Talos OS or Kubernetes. Use to upgrade talos, bump kubernetes version, edit talenv, run a rolling upgrade across nodes, or change the talos version. High-stakes — one node at a time, etcd quorum gated.

Use when Codex needs to inspect or troubleshoot the Anton cluster's self-hosted Temporal deployment with the local Temporal CLI, including checking cluster health, namespaces, workflow visibility, schedules, search attributes, Web UI reachability, or Kubernetes readiness for the `temporal` namespace. Prefer this for Temporal CLI tasks in Anton rather than generic Temporal SDK guidance.

Remote access reference for Anton. Use when Codex needs to reach or reason about kubectl, flux, talosctl, Tailscale MagicDNS, kubeconfig, talosconfig, off-LAN node access, or failed connectivity to k8s nodes.

Reference for Anton Flux and manifest conventions. Use when Codex writes or reviews Kubernetes manifests, adds a Flux app, chooses OCIRepository versus HelmRepository versus GitRepository, decides between ExternalSecret and SOPS, or checks the 3-file app pattern.

Ordered Flux reconciliation triage for Anton. Use when a Kustomization, HelmRelease, GitRepository, OCIRepository, HelmRepository, SOPS decryption, postBuild substitution, dependency, or Flux sync is stuck or not progressing.

Read-only Talos OS triage for Anton. Use to check cluster health, node readiness, etcd quorum, Talos services, disks, NVMe inventory, install disks, routes, interfaces, MTU, VIP, inter-node reachability, DNS, or suspected node-level failures.

Triage ntfy.sh-routed alerts in Anton — identify which alert fired, why it fired (or why it didn't deliver), and propose a fix. Use when "got an ntfy alert", "alert just fired", "ntfy not delivering", "AlertmanagerClusterFailedToSendAlerts", "AlertmanagerFailedToSendAlerts", "code 40014", "attachments not allowed", "iOS push missing", "test the ntfy receiver", "send a test alert", "what just paged me", "is ntfy working". Combines kube-prometheus-stack Alertmanager API, the self-hosted ntfy server (ADR 0026), and the ntfy CLI for poll/publish probes. Read-only by default; proposes edits the operator applies.

Intake gate for adding new system or infrastructure components to Anton. Asks the user to declare intent (concrete need, honest learning, or both), then applies the matching rubric — full production rubric for concrete need, contained-learning rubric for learning intake — and returns add / defer / reject with an ADR-ready summary. Welcomes honest learning intake (anton is partly a learning cluster; "things that don't scale" are okay when declared) but rejects completionism dressed as need. Read-only — never scaffolds manifests, never applies to the cluster. Use when asking "should I add X", "can I run X on the cluster", "is X worth adopting", "I want to try X", "I want to learn X", "evaluate new component", "vet this helm chart", "cluster intake", "new app decision", before scaffolding a new Flux app, or when tempted by a shiny project on HN. Hands passing candidates off to add-flux-app. Keywords — intake, adopt, install, new component, new app, evaluate, should I run, worth it, learning, experiment, try out,

Triage a flux sync stuck in Anton. Ordered diagnostic path when flux is stuck, kustomization not progressing, HelmRelease failing, SOPS decryption error blocks sync, or dependency not ready. Force reconcile in safe dependency order.

Expose a workload for access. Four paths: envoy-internal (LAN via split-horizon DNS), Tailscale Ingress (internal remote HTTP with browser-trusted TLS), Tailscale Service annotation (raw TCP / non-HTTP), envoy-external + Cloudflare tunnel (genuinely public, requires explicit approval). Handles HTTPRoute authoring, DNSEndpoint for secondary domains, and per-domain cert wiring.

Reference for Anton's Flux conventions — the 3-file pattern, ks.yaml shape, HelmRelease sources (OCI/Helm/Git), postBuild substitution from cluster-secrets, and SOPS vs ExternalSecret decision. Use when writing manifests or picking a secret store.

Anton planner skill — author, update, and close multi-session initiatives (migrations, rollouts, long-running refactors) in `context/plans/`. Use when starting a multi-session initiative, tracking next steps on in-flight work, migrating a memory entry to a durable plan, closing a completed initiative, or reviewing what's open. Plans live at `context/plans/NNNN-kebab-slug.md` and are mutable — they capture execution state (what's next, what's blocked, log of decisions made during work) while ADRs capture immutable decisions (why). The active-plan index is built by scanning plan files directly and injected into every Claude Code session by `.claude/hooks/inject_plans_index.py`. Keywords — plan, planner, initiative, track work, multi-session, next steps, checklist, migration plan, rollout plan, roadmap, in-flight, blocker, close plan, review-by, exit plan, timebox, memory-to-plan handoff.

Anton ADR lifecycle — author new architectural decision records, list existing ones by status or affects-category, and mark old decisions superseded. Use when capturing a decision (especially after `cluster-intake-gatekeeper` returns an ADD/DEFER/REJECT verdict), when reviewing prior decisions before changing direction, when checking if a candidate component has been removed before, or when promoting a decision out of memory into a durable record. ADRs live in `context/adrs/` and are immutable — supersession is the only way to change a decision. The ADR index is built by scanning ADR files directly and injected into every Claude Code session by `.claude/hooks/inject_adr_index.py`. Keywords — ADR, architecture decision record, decision log, supersede, decision history, why did we, prior decision, recorded decision, MADR, immutable, intake handoff, cluster-intake-gatekeeper handoff, removal graveyard, reverted decision.

Kubernetes-layer health triage for Anton. Starts above the Talos OS layer and walks CNI/DNS → Flux → platform components (cert-manager, ESO, envoy-gateway, cloudflared) → apps. Use when asking "is the cluster ok", "anything broken", "is Flux healthy", "platform components healthy", "ESO stuck", "gateway not serving", "cloudflared tunnel disconnected", "cert not issued". For node, disk, etcd, or network issues, hand off to talos-inspect first.

Remote access reference for Anton — how to reach the cluster from off-LAN via talosctl, kubectl, and Tailscale MagicDNS. Use when working with talosconfig, kubectl context, tailscale, remote access, or when you cannot reach k8s nodes.

Read-only triage skill for Anton's Talos OS layer. Use to check cluster health, inspect node disks (install disks, serials, NVMe, volumes), or diagnose node networking (interfaces, routes, bonds, VIP, inter-node reachability, DNS, MTU). Keywords — talos health, cluster healthy, healthcheck, check nodes, node triage, etcd members, etcd quorum, talosctl get disks, NVMe, install disk, disk inventory, volume status, talos network, interface, link, route, bond, VIP, MTU, duplicate IP, resolvers, clock skew.