[CONTP-2086][CONTP-2085] Expose cluster_checks metrics via COAT

comp/core/agenttelemetry/impl/agenttelemetry_test.go

@@ -2982,6 +2982,52 @@ func TestInstrumentationControllerMetricsInClusterAgentProfile(t *testing.T) {
 	assert.ElementsMatch(t, []string{"section", "status"}, metrics["instrumentation_controller.reconciliations"])
 }
 
+// TestClusterChecksMetricsInClusterAgentProfile guards the cluster-checks / CLC telemetry
+// onboarded to COAT. A metric missing from the allowlist is dropped rather than shipped.
+//
+// These metrics are intentionally emitter-only: their only distinguishing label is the
+// per-node (CLC runner) label, which is high-cardinality and must be dropped, so COAT sums
+// them into meaningful cluster-wide counts/totals (configs dispatched, nodes reporting,
+// rebalancing decisions/moves, failed stats collections). busyness and predicted_utilization
+// are deliberately NOT onboarded: they are per-runner values (a ratio, for utilization) whose
+// additive aggregation across runners is meaningless, so they must stay out of this profile.
+func TestClusterChecksMetricsInClusterAgentProfile(t *testing.T) {
+	cfg, err := parseConfig(configmock.NewFromYAML(t, defaultProfiles))
+	require.NoError(t, err)
+
+	var profile *Profile
+	for _, candidate := range cfg.Profiles {
+		if candidate.Name == "cluster-agent" {
+			profile = candidate
+			break
+		}
+	}
+	require.NotNil(t, profile)
+	require.NotNil(t, profile.Metric)
+
+	metrics := make(map[string][]string, len(profile.Metric.Metrics))
+	for _, metric := range profile.Metric.Metrics {
+		metrics[metric.Name] = metric.PreserveTags
+	}
+
+	// Onboarded, emitter-only (no preserve_tags).
+	for _, name := range []string{
+		"cluster_checks.nodes_reporting",
+		"endpoint_checks.configs_dispatched",
+		"cluster_checks.rebalancing_decisions",
+		"cluster_checks.successful_rebalancing_moves",
+		"cluster_checks.failed_stats_collection",
+	} {
+		preserved, ok := metrics[name]
+		require.Truef(t, ok, "%s is not allowlisted, so it would never be sent", name)
+		assert.Emptyf(t, preserved, "%s must be emitter-only (no preserve_tags)", name)
+	}
+
+	// Deliberately excluded: per-runner values whose additive aggregation is meaningless.
+	assert.NotContains(t, metrics, "cluster_checks.busyness")
+	assert.NotContains(t, metrics, "cluster_checks.predicted_utilization")
+}
+
 // TestDataPlanePreflightModeProfile guards the Agent Data Plane preflight mode metrics.
 //
 // The pre-flight in comp/dataplane/preflightmode reports its outcome purely through these

comp/core/agenttelemetry/impl/defaultProfiles.yaml

@@ -456,6 +456,20 @@ profiles:
       - name: cluster_checks.unscheduled_check
         preserve_tags:
           - config_source
+      # cluster checks: node reporting and endpoint-check dispatch. These carry only
+      # a per-node label (dropped, high-cardinality) so they aggregate to emitter-only
+      # counts/totals.
+      # Accepted limitation: zero_metric exclusion (below) drops nodes_reporting==0, so the
+      # "no runners reporting" state produces no signal. Only positive counts are observable.
+      - name: cluster_checks.nodes_reporting
+      - name: endpoint_checks.configs_dispatched
+      # cluster checks: rebalancing activity (additive counters). Sustained high rates
+      # indicate an unstable or churning check distribution.
+      - name: cluster_checks.rebalancing_decisions
+      - name: cluster_checks.successful_rebalancing_moves
+      # cluster checks: stats-collection health. failed_stats_collection is a direct
+      # error signal (DCA cannot collect stats from a runner).
+      - name: cluster_checks.failed_stats_collection
       - name: language_detection_patcher.patches
         preserve_tags:
           - owner_kind

releasenotes/notes/clusterchecks-coat-telemetry-ff5b5929160cdfb8.yaml

@@ -0,0 +1,14 @@
+---
+other:
+  - |
+    Added Cross-Org Agent Telemetry (COAT) metrics to the ``cluster-agent`` profile to
+    surface cluster checks and Cluster Level Check (CLC) runner health fleet-wide. The
+    metrics are emitter-only (the per-node label is aggregated away):
+
+    | Metric Name | Type | Description |
+    | --- | --- | --- |
+    | ``cluster_checks.nodes_reporting`` | Gauge | Number of node agents reporting to the Cluster Agent |
+    | ``endpoint_checks.configs_dispatched`` | Gauge | Number of endpoint check configurations dispatched |
+    | ``cluster_checks.rebalancing_decisions`` | Counter | Total number of check rebalancing decisions |
+    | ``cluster_checks.successful_rebalancing_moves`` | Counter | Total number of successful check rebalancing moves |
+    | ``cluster_checks.failed_stats_collection`` | Counter | Total number of unsuccessful runner stats collection attempts |