メニュー
Use when the user asks to "ver servicios", "check services", "pod status", "estado del cluster", "qué está corriendo", "ver pods", or "cluster status" in the Afianza Kubernetes clusters. SKIP for log analysis (use ops-suite:service-logs) or database queries.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Use when the user wants to release, tag, publish, or deploy a new version to dev. Triggers on "release", "tag", "nueva versión", "bump version", "subir versión", "publicar versión", "deploy dev", "sacar release", or any mention of creating a v* tag. Use this skill whenever the user wants to cut a new version of any Afianza service, even if they don't say "release" explicitly.
Stress-test a feature spec by generating use cases from normal to extreme, one at a time interactively. Use when reviewing a spec for gaps, undefined behavior, or missing edge cases — even if the user just says "review this spec", "find holes in this", or "check my ACs".
Use the moment a technical implementation plan has been generated — by writing-plans, native plan mode / ExitPlanMode, or hand-written — and BEFORE presenting it to the user for review, approval, or implementation. The gate sits between "plan written" and "plan shown/built". NOT for a brainstorming spec (a spec guides the plan; it is not yet a technical plan, so there is nothing to gate). Triggers on a freshly produced plan, plan handoff, "here is the plan", pre-review / pre-implementation rule check.
Use when connecting to an Afianza service database for inspection, debugging, or ad-hoc queries. Covers local, dev, and prod environments for all services (pgi, pc, pd). Use before running psql or querying any service's PostgreSQL database.
Use when the user wants to deploy to production, "subir a prod", "desplegar en producción", "lanzar a prod", "manual deploy", "deploy prod", or promote a specific version/tag to the production environment. Use for any Afianza service. SKIP if the user only wants to release/tag (use the `release` skill instead).
SOC 職業分類に基づく
| name | k8s-services |
| description | Use when the user asks to "ver servicios", "check services", "pod status", "estado del cluster", "qué está corriendo", "ver pods", or "cluster status" in the Afianza Kubernetes clusters. SKIP for log analysis (use ops-suite:service-logs) or database queries. |
| allowed-tools | Bash AskUserQuestion |
| metadata | {"argument-hint":"[namespace] [service-name]"} |
| Context | Cluster | Namespaces |
|---|---|---|
k8s-afianza-dev | Azure dev (new) | pgi, plataforma-datos, portal-cliente, broker |
dev | vCluster dev (old) | plataformadato |
afianza-prod | Production | pgi, plataforma-datos, portal-cliente, broker |
If $ARGUMENTS contains a context or environment name, use it. Otherwise default to current context (kubectl config current-context).
If $ARGUMENTS contains a namespace or service name, scope to it. Otherwise show all relevant namespaces.
Run in order:
# Resolve context (use $ARGUMENTS value if specified, else current)
KCTX=$(kubectl config current-context)
# Pods across relevant namespaces
kubectl --context "$KCTX" get pods -n pgi
kubectl --context "$KCTX" get pods -n plataforma-datos
kubectl --context "$KCTX" get pods -n portal-cliente
kubectl --context "$KCTX" get pods -n broker
If a specific namespace or service was requested, only run that query.
Identify pods that are NOT in Running or Completed state:
CrashLoopBackOff — application error, suggest checking logs with ops-suite:service-logsImagePullBackOff — wrong image tag or registry auth issuePending — scheduling problem (node resources or tolerations)OOMKilled — memory limit exceededContext: k8s-afianza-dev
namespace: pgi
NAME READY STATUS RESTARTS AGE
pgi-service-pgi-api-xxxx-yyyy 1/1 Running 0 2d
...
namespace: plataforma-datos
...
Issues:
- pod X in namespace Y: CrashLoopBackOff — run /ops-suite:service-logs X dev
If no issues found, print "All pods healthy ✓".