feat: close preprod observability loop

docs/TASKS/observability/003-preprod-operations-loop.md

@@ -0,0 +1,34 @@
+# Observability 003: Preprod Operations Loop
+
+## Goal
+
+Close the preproduction operations loop by adding alert delivery scaffolding, uptime probes, workflow health gauges, secured Grafana guidance, and an operator runbook.
+
+## Feature Spec
+
+- `docs/FEATURES/observability.md`
+
+## Scope
+
+- Add Alertmanager to the optional observability compose overlay.
+- Add Blackbox Exporter uptime probes for the web container and API readiness endpoint.
+- Add backend database-derived workflow health gauges.
+- Add Prometheus alerts for uptime probes and workflow health.
+- Add an optional Caddy snippet for protected Grafana exposure.
+- Add an operator runbook for bring-up, alert triage, and security defaults.
+
+## Out Of Scope
+
+- Operating the remote preproduction host.
+- Choosing the final alert destination.
+- Client-facing status page.
+- External third-party uptime monitoring.
+
+## Validation
+
+```bash
+dotnet build backend/Socialize.slnx
+dotnet test backend/Socialize.slnx
+docker compose -f deploy/compose.yml -f deploy/observability/compose.observability.yml config
+jq empty deploy/observability/grafana/dashboards/socialize-overview.json
+```