feat(observability): service discovery, health cards, alertmanager widget

Slice 3 of observability-service-registry (frontend). The Observability
page discovers Grafana from the service registry instead of env vars,
adds Grafana + Prometheus health cards, and ships an alertmanager
active_alerts dashboard widget.

- types: added GrafanaStatus + PrometheusStatus; added optional
  service_id/error to AlertmanagerStatus.
- api/client.ts + hooks/useObservability.ts: fetchGrafanaStatus,
  fetchPrometheusStatus, useGrafanaStatus, usePrometheusStatus.
- widgets/AlertmanagerAlertsWidget.tsx (new): presentational widget
  consuming the active_alerts summary shape (total/by_severity/alerts);
  exported from widgets/index.ts.
- integrations/registry.ts: alertmanager binding (active_alerts kind,
  30s refresh, optional severity_filter); registry.test.ts updated to
  6 service types incl alertmanager + a resolve test.
- components/ObservabilityPage.tsx: removed
  import.meta.env.VITE_GRAFANA_URL; derive GRAFANA_BASE_URL from the
  first enabled grafana service via useServiceInstances("grafana");
  added Grafana + Prometheus HealthCards (up/not-configured/unreachable)
  with QueryError retry blocks; machine dashboard shows a "No Grafana
  service configured" empty-state linking to /services when none is set.

npm run build (tsc -b + vite) clean; 0 lint errors; 72 frontend tests
pass. Reviewed fresh-context (read-only): no blockers.

backend/src/media_library_viewer_api/routers/monitoring.py

@@ -56,9 +56,7 @@ def _auth_headers(service: ServiceRecord) -> dict[str, str]:
     return {"Authorization": f"Bearer {api_key}"} if api_key else {}
 
 
-def _status_response(
+def _status_response(service: ServiceRecord | None, *, version: str = "", error: str | None = None) -> dict[str, Any]:
-    service: ServiceRecord | None, *, version: str = "", error: str | None = None
-) -> dict[str, Any]:
     return {
         "up": error is None,
         "version": version or "",
@@ -232,9 +230,7 @@ def get_prometheus_status(
     try:
         health = requests.get(f"{base}/-/healthy", headers=headers, timeout=timeout)
         health.raise_for_status()
-        build_info = requests.get(
+        build_info = requests.get(f"{base}/api/v1/status/buildinfo", headers=headers, timeout=timeout)
-            f"{base}/api/v1/status/buildinfo", headers=headers, timeout=timeout
-        )
         build_info.raise_for_status()
         version = build_info.json().get("data", {}).get("version", "")
     except Exception:

backend/tests/test_api.py

@@ -636,6 +636,7 @@ class TestResolveServiceRecord:
     def _store(self, rows):
         store = MagicMock()
         store.get_service = lambda sid: next((r for r in rows if r["id"] == sid), None)
+
         def list_filtered(service_type=None):
             return [r for r in rows if r["service_type"] == (service_type or r["service_type"])]
 
@@ -882,7 +883,9 @@ class TestGrafanaStatus:
         assert data["error"] == "no_service_configured"
 
     def test_grafana_status_when_unreachable(self, test_client):
-        service = ServiceRecord(id="g1", service_type="grafana", name="Grafana", config={"base_url": "http://grafana:3000"})
+        service = ServiceRecord(
+            id="g1", service_type="grafana", name="Grafana", config={"base_url": "http://grafana:3000"}
+        )
         with (
             patch(f"{_MON}._resolve_service_record", return_value=service),
             patch(f"{_MON}.requests.get", side_effect=Exception("refused")),
@@ -895,7 +898,9 @@ class TestGrafanaStatus:
         assert data["name"] == "Grafana"
 
     def test_grafana_status_returns_version(self, test_client):
-        service = ServiceRecord(id="g1", service_type="grafana", name="Grafana", config={"base_url": "http://grafana:3000"})
+        service = ServiceRecord(
+            id="g1", service_type="grafana", name="Grafana", config={"base_url": "http://grafana:3000"}
+        )
         resp = MagicMock()
         resp.json.return_value = {"version": "11.3.1", "database": "ok"}
         resp.raise_for_status = MagicMock()
@@ -920,7 +925,9 @@ class TestPrometheusStatus:
         assert data["error"] == "no_service_configured"
 
     def test_prometheus_status_when_unreachable(self, test_client):
-        service = ServiceRecord(id="p1", service_type="prometheus", name="Prometheus", config={"base_url": "http://prometheus:9090"})
+        service = ServiceRecord(
+            id="p1", service_type="prometheus", name="Prometheus", config={"base_url": "http://prometheus:9090"}
+        )
         with (
             patch(f"{_MON}._resolve_service_record", return_value=service),
             patch(f"{_MON}.requests.get", side_effect=Exception("refused")),
@@ -932,7 +939,9 @@ class TestPrometheusStatus:
         assert data["error"] == "prometheus_unreachable"
 
     def test_prometheus_status_returns_version(self, test_client):
-        service = ServiceRecord(id="p1", service_type="prometheus", name="Prometheus", config={"base_url": "http://prometheus:9090"})
+        service = ServiceRecord(
+            id="p1", service_type="prometheus", name="Prometheus", config={"base_url": "http://prometheus:9090"}
+        )
         health = MagicMock()
         health.raise_for_status = MagicMock()
         build_info = MagicMock()

frontend/src/api/client.ts

@@ -33,6 +33,8 @@ import type {
 	DashboardShortcutInput,
 	AlertmanagerAlertSummary,
 	AlertmanagerStatus,
+	GrafanaStatus,
+	PrometheusStatus,
 	PrometheusTarget,
 } from "../types";
 
@@ -387,5 +389,11 @@ export const fetchAlertmanagerAlerts = () =>
 export const fetchAlertmanagerStatus = () =>
 	get<AlertmanagerStatus>("/api/monitoring/alertmanager-status");
 
+export const fetchGrafanaStatus = () =>
+	get<GrafanaStatus>("/api/monitoring/grafana-status");
+
+export const fetchPrometheusStatus = () =>
+	get<PrometheusStatus>("/api/monitoring/prometheus-status");
+
 export const fetchPrometheusTargets = () =>
 	get<PrometheusTarget[]>("/api/monitoring/prometheus-targets");

frontend/src/components/ObservabilityPage.tsx

@@ -7,6 +7,7 @@ import {
 	CheckCircle2,
 	ChevronDown,
 	ExternalLink,
+	Gauge,
 	Inbox,
 	Radio,
 	RefreshCw,
@@ -17,9 +18,12 @@ import {
 import {
 	useAlertmanagerAlerts,
 	useAlertmanagerStatus,
+	useGrafanaStatus,
+	usePrometheusStatus,
 	usePrometheusTargets,
 	useMonitoringMachines,
 } from "../hooks/useObservability";
+import { useServiceInstances } from "../hooks/useServices";
 import { Card, CardContent, CardHeader, CardTitle } from "@/components/ui/card";
 import { Badge } from "@/components/ui/badge";
 import { Alert, AlertDescription, AlertTitle } from "@/components/ui/alert";
@@ -43,9 +47,6 @@ import type {
 	PrometheusTarget,
 } from "../types";
 
-const GRAFANA_BASE_URL =
-	import.meta.env.VITE_GRAFANA_URL || "http://localhost:3000";
-
 function severityVariant(
 	severity: string,
 ): "default" | "secondary" | "destructive" | "outline" {
@@ -283,6 +284,18 @@ export function ObservabilityPage() {
 		error: statusError,
 		refetch: refetchStatus,
 	} = useAlertmanagerStatus();
+	const {
+		data: grafanaStatus,
+		isLoading: grafanaLoading,
+		error: grafanaError,
+		refetch: refetchGrafana,
+	} = useGrafanaStatus();
+	const {
+		data: prometheusStatus,
+		isLoading: prometheusLoading,
+		error: prometheusError,
+		refetch: refetchPrometheus,
+	} = usePrometheusStatus();
 	const {
 		data: prometheusTargets,
 		isLoading: targetsLoading,
@@ -295,8 +308,14 @@ export function ObservabilityPage() {
 		error: machinesError,
 		refetch: refetchMachines,
 	} = useMonitoringMachines();
+	const { data: grafanaServices = [] } = useServiceInstances("grafana");
 	const [selectedMachineId, setSelectedMachineId] = useState<string>("");
 
+	const grafanaService =
+		grafanaServices.find((s) => s.enabled) ?? grafanaServices[0];
+	const GRAFANA_BASE_URL =
+		(grafanaService?.config?.base_url as string | undefined) ?? "";
+
 	const selectedMachine = useMemo<MonitoringMachine | null>(
 		() =>
 			machines.find((m) => m.id === selectedMachineId) ?? machines[0] ?? null,
@@ -304,13 +323,13 @@ export function ObservabilityPage() {
 	);
 
 	const nodeExporterDashboardUrl = useMemo(() => {
-		if (!selectedMachine) return "";
+		if (!selectedMachine || !GRAFANA_BASE_URL) return "";
 		const instance = `${selectedMachine.host || "localhost"}:9100`;
 		return `${GRAFANA_BASE_URL}/d/node-exporter-overview/node-exporter-overview?kiosk&var-instance=${encodeURIComponent(instance)}`;
-	}, [selectedMachine]);
+	}, [selectedMachine, GRAFANA_BASE_URL]);
 
 	const logsUrl = useMemo(() => {
-		if (!selectedMachine) return "";
+		if (!selectedMachine || !GRAFANA_BASE_URL) return "";
 		const container =
 			selectedMachine.mode === "local" ? "backend" : selectedMachine.name;
 		return `${GRAFANA_BASE_URL}/explore?orgId=1&left=${encodeURIComponent(
@@ -320,7 +339,7 @@ export function ObservabilityPage() {
 				range: { from: "now-1h", to: "now" },
 			}),
 		)}`;
-	}, [selectedMachine]);
+	}, [selectedMachine, GRAFANA_BASE_URL]);
 
 	const alertmanagerStatusDetail = alertmanagerStatus?.up
 		? alertmanagerStatus.version
@@ -402,6 +421,52 @@ export function ObservabilityPage() {
 					icon={Server}
 					isLoading={machinesLoading}
 				/>
+				<HealthCard
+					title="Grafana"
+					status={
+						grafanaError
+							? "error"
+							: grafanaStatus?.up
+								? "ok"
+								: grafanaLoading
+									? "unknown"
+									: "error"
+					}
+					detail={
+						grafanaStatus?.up
+							? grafanaStatus.version
+								? `version ${grafanaStatus.version}`
+								: "reachable"
+							: grafanaStatus?.error === "no_service_configured"
+								? "not configured"
+								: "unreachable"
+					}
+					icon={Gauge}
+					isLoading={grafanaLoading}
+				/>
+				<HealthCard
+					title="Prometheus"
+					status={
+						prometheusError
+							? "error"
+							: prometheusStatus?.up
+								? "ok"
+								: prometheusLoading
+									? "unknown"
+									: "error"
+					}
+					detail={
+						prometheusStatus?.up
+							? prometheusStatus.version
+								? `version ${prometheusStatus.version}`
+								: "reachable"
+							: prometheusStatus?.error === "no_service_configured"
+								? "not configured"
+								: "unreachable"
+					}
+					icon={Radio}
+					isLoading={prometheusLoading}
+				/>
 			</div>
 
 			<div className="space-y-3">
@@ -433,6 +498,20 @@ export function ObservabilityPage() {
 						refetch={refetchMachines}
 					/>
 				)}
+				{grafanaError && (
+					<QueryError
+						label="Grafana status"
+						error={grafanaError}
+						refetch={refetchGrafana}
+					/>
+				)}
+				{prometheusError && (
+					<QueryError
+						label="Prometheus status"
+						error={prometheusError}
+						refetch={refetchPrometheus}
+					/>
+				)}
 			</div>
 
 			{alertsSummary?.error && (
@@ -542,7 +621,8 @@ export function ObservabilityPage() {
 						</Select>
 					</CardHeader>
 					<CardContent className="space-y-4">
-						{selectedMachine ? (
+					{selectedMachine ? (
+						GRAFANA_BASE_URL ? (
 							<>
 								<GrafanaLinkCard
 									title={`${selectedMachine.name} metrics`}
@@ -556,6 +636,18 @@ export function ObservabilityPage() {
 								/>
 							</>
 						) : (
+							<EmptyState
+								icon={Gauge}
+								title="No Grafana service configured"
+								description="Add a Grafana service instance to enable deep-links to dashboards and logs."
+								action={
+									<Button variant="outline" size="sm" asChild>
+										<Link to="/services">Open Services</Link>
+									</Button>
+								}
+							/>
+						)
+					) : (
 							<EmptyState
 								icon={ServerOff}
 								title="No machine selected"

frontend/src/hooks/useObservability.ts

@@ -2,6 +2,8 @@ import { useQuery } from "@tanstack/react-query";
 import {
 	fetchAlertmanagerAlerts,
 	fetchAlertmanagerStatus,
+	fetchGrafanaStatus,
+	fetchPrometheusStatus,
 	fetchPrometheusTargets,
 	fetchMonitoringMachines,
 } from "../api/client";
@@ -26,6 +28,26 @@ export function useAlertmanagerStatus() {
 	});
 }
 
+export function useGrafanaStatus() {
+	return useQuery({
+		queryKey: ["observability", "grafana-status"],
+		queryFn: fetchGrafanaStatus,
+		retry: 2,
+		staleTime: 10_000,
+		refetchInterval: 30_000,
+	});
+}
+
+export function usePrometheusStatus() {
+	return useQuery({
+		queryKey: ["observability", "prometheus-status"],
+		queryFn: fetchPrometheusStatus,
+		retry: 2,
+		staleTime: 10_000,
+		refetchInterval: 30_000,
+	});
+}
+
 export function usePrometheusTargets() {
 	return useQuery({
 		queryKey: ["observability", "prometheus-targets"],

frontend/src/integrations/registry.test.ts

@@ -9,8 +9,9 @@ import {
 import type { ServiceInstance, WidgetInstance } from "../types";
 
 describe("service registry", () => {
-	it("registers the five backend service types", () => {
+	it("registers the backend service types", () => {
 		expect(Object.keys(SERVICE_REGISTRY).sort()).toEqual([
+			"alertmanager",
 			"grafana",
 			"jellyfin",
 			"nextcloud",
@@ -23,6 +24,9 @@ describe("service registry", () => {
 		expect(SERVICE_REGISTRY.grafana.widgets.map((w) => w.kind)).toEqual([
 			"link",
 		]);
+		expect(SERVICE_REGISTRY.alertmanager.widgets.map((w) => w.kind)).toEqual([
+			"active_alerts",
+		]);
 		expect(SERVICE_REGISTRY.ssh_tasks.widgets.map((w) => w.kind)).toEqual([
 			"task_output",
 		]);
@@ -62,6 +66,35 @@ describe("service registry", () => {
 		expect(resolved?.refreshIntervalMs).toBe(0);
 	});
 
+	it("resolves an alertmanager active_alerts widget", () => {
+		const widget: WidgetInstance = {
+			id: "wam",
+			service_id: "sam",
+			widget_kind: "active_alerts",
+			title: "Alerts",
+			config: {},
+			enabled: true,
+			sort_order: 0,
+			created_at: 0,
+			updated_at: 0,
+		};
+		const services: ServiceInstance[] = [
+			{
+				id: "sam",
+				service_type: "alertmanager",
+				name: "Alertmanager",
+				config: { base_url: "http://am:9093" },
+				secrets_set: {},
+				enabled: true,
+				created_at: 0,
+				updated_at: 0,
+			},
+		];
+		const resolved = resolveWidget(widget, services);
+		expect(resolved).toBeDefined();
+		expect(resolved?.refreshIntervalMs).toBe(30_000);
+	});
+
 	it("resolves a built-in widget without a service", () => {
 		const widget: WidgetInstance = {
 			id: "w2",

frontend/src/integrations/registry.ts

@@ -1,4 +1,5 @@
 import type { ComponentType } from "react";
+import { AlertmanagerAlertsWidget } from "../widgets/AlertmanagerAlertsWidget";
 import { BackupsWidget } from "../widgets/BackupsWidget";
 import { GrafanaLinkWidget } from "../widgets/GrafanaLinkWidget";
 import { JellyfinWidget } from "../widgets/JellyfinWidget";
@@ -42,6 +43,28 @@ export interface ServiceBinding {
 }
 
 export const SERVICE_REGISTRY: Record<string, ServiceBinding> = {
+	alertmanager: {
+		serviceType: "alertmanager",
+		name: "Alertmanager",
+		description: "Alertmanager alerts and status.",
+		widgets: [
+			{
+				kind: "active_alerts",
+				name: "Active alerts",
+				description: "Firing alerts summary from Alertmanager.",
+				refreshIntervalMs: 30_000,
+				defaultConfig: {},
+				configSchema: {
+					type: "object",
+					properties: {
+						severity_filter: { type: "string" },
+					},
+					required: [],
+				},
+				component: AlertmanagerAlertsWidget,
+			},
+		],
+	},
 	grafana: {
 		serviceType: "grafana",
 		name: "Grafana",

frontend/src/types/index.ts

@@ -386,6 +386,24 @@ export interface AlertmanagerStatus {
 	uptime: string;
 	name: string;
 	peers: string[];
+	service_id?: string;
+	error?: string | null;
+}
+
+export interface GrafanaStatus {
+	up: boolean;
+	version: string;
+	service_id: string;
+	name: string;
+	error?: string | null;
+}
+
+export interface PrometheusStatus {
+	up: boolean;
+	version: string;
+	service_id: string;
+	name: string;
+	error?: string | null;
 }
 
 export interface PrometheusTarget {

frontend/src/widgets/AlertmanagerAlertsWidget.tsx

@@ -0,0 +1,93 @@
+import { Alert, AlertDescription } from "@/components/ui/alert";
+import { Badge } from "@/components/ui/badge";
+import { Skeleton } from "@/components/ui/skeleton";
+import { SectionCard } from "../components/SectionCard";
+import { useWidgetData } from "../hooks/useWidgets";
+import type {
+	AlertmanagerAlert,
+	AlertmanagerAlertSummary,
+	WidgetInstance,
+} from "../types";
+
+interface Props {
+	widget: WidgetInstance;
+	refreshIntervalMs: number;
+	description?: string;
+}
+
+function severityVariant(
+	severity: string,
+): "default" | "secondary" | "destructive" | "outline" {
+	switch (severity.toLowerCase()) {
+		case "critical":
+			return "destructive";
+		case "warning":
+			return "default";
+		case "info":
+			return "secondary";
+		default:
+			return "outline";
+	}
+}
+
+export function AlertmanagerAlertsWidget({
+	widget,
+	refreshIntervalMs,
+	description,
+}: Props) {
+	const { data, isLoading } = useWidgetData(widget.id, refreshIntervalMs);
+	const summary = data?.data as AlertmanagerAlertSummary | undefined;
+	const alerts: AlertmanagerAlert[] = summary?.alerts ?? [];
+
+	return (
+		<SectionCard title={widget.title} description={description}>
+			{isLoading && !data ? (
+				<div className="space-y-2">
+					<Skeleton className="h-8 w-full" />
+					<Skeleton className="h-8 w-full" />
+				</div>
+			) : data?.error ? (
+				<Alert variant="destructive">
+					<AlertDescription>{data.error}</AlertDescription>
+				</Alert>
+			) : summary ? (
+				<div className="space-y-3">
+					<div className="flex items-center gap-3">
+						<div>
+							<div className="text-2xl font-semibold">{summary.total}</div>
+							<div className="text-xs text-muted-foreground">
+								Firing alert{summary.total === 1 ? "" : "s"}
+							</div>
+						</div>
+						<div className="flex flex-wrap gap-1">
+							{Object.entries(summary.by_severity).map(([sev, count]) => (
+								<Badge key={sev} variant={severityVariant(sev)}>
+									{sev}: {count}
+								</Badge>
+							))}
+						</div>
+					</div>
+					{alerts.length > 0 ? (
+						<ul className="space-y-1.5">
+							{alerts.slice(0, 5).map((a, idx) => (
+								<li
+									key={`${a.name}-${idx}`}
+									className="flex items-center justify-between gap-2 rounded-md border px-2 py-1 text-sm"
+								>
+									<span className="truncate">{a.name}</span>
+									<Badge variant={severityVariant(a.severity)}>
+										{a.severity}
+									</Badge>
+								</li>
+							))}
+						</ul>
+					) : (
+						<div className="text-xs text-muted-foreground">
+							Everything looks quiet.
+						</div>
+					)}
+				</div>
+			) : null}
+		</SectionCard>
+	);
+}

frontend/src/widgets/index.ts

@@ -1,3 +1,4 @@
+export { AlertmanagerAlertsWidget } from "./AlertmanagerAlertsWidget";
 export { BackupsWidget } from "./BackupsWidget";
 export { GrafanaLinkWidget } from "./GrafanaLinkWidget";
 export { JellyfinWidget } from "./JellyfinWidget";