metrics-graphanalisted

# Very Happy monitoring Read [docs/monitoring/README.md](../../../docs/monitoring/README.md) first: it owns the deployed topology, object names, data semantics and rollback. Application release remains owned by the `release` skill and `docs/operations.md`. ## Choose the actual management path - Production uses the existing sy Grafana/Prometheus stack. Reach its Kubernetes API through `ssh k8s`; verify host identity and current Service addresses before queries. A local kubeconfig context named sy is not evidence its credentials work. - The connection dashboard is managed by the dedicated ConfigMap documented in the runbook. Edit `docs/monitoring/connection-quality.dashboard.json`, resolve `${DS_PROMETHEUS}` to the verified existing data source UID when provisioning, and change only the owned ConfigMap after exporting its previous state. - Its existing provider has `allowUiUpdates=false`. Do not switch to an API/UI overwrite, change the global provider, or install grafanactl just to update it. Other dashboards may have a different owner; inspect before changing them. - Do not assume a repo `.env`, Basic Auth, or an initial Kubernetes admin Secret provides current Grafana access. An auth failure is not permission to reset a password. Never print credentials or persist session cookies in artifacts. ## Query and verify Resolve the Prometheus Service address using the cluster API, then query its `/api/v1/query` or `/api/v1/query_range` from the authorized pri