diff --git a/docs/scheduler/observability.md b/docs/scheduler/observability.md
index 59abf7a8..ce560205 100644
--- a/docs/scheduler/observability.md
+++ b/docs/scheduler/observability.md
@@ -208,8 +208,8 @@ The swarm's collector adds `hive`, the name of the hive whose receiver accepted
 the push (see [Authenticated ingest](#authenticated-ingest)). The container
 doesn't set it.
 
-Append additional labels via `extraResourceAttributes` (see option
-reference above); pass custom per-data-point labels with
+Append additional swarm-wide labels via `extraResourceAttributes` (see
+option reference above); pass custom per-data-point labels with
 `hive-metric --labels` (see below).
 
 ## Log severity
@@ -322,9 +322,10 @@ exists, so off-the-shelf OTEL/Grafana container dashboards work. Resource
 
 The `hyperhive.`-prefixed metrics have no semconv equivalent (memory
 limit + peak, on-disk footprint, and an instantaneous cpu percent kept
-alongside the spec `container.cpu.time` counter for convenience). Hive
-labels (`hive`, `swarm`, …) ride on the resource via
-`extraResourceAttributes`.
+alongside the spec `container.cpu.time` counter for convenience). The `swarm`
+label rides on the resource, and the swarm collector adds `hive` from the
+receiver that accepted the sample. `extraResourceAttributes` is swarm-wide, so
+it can't carry a per-hive label.
 
 <!-- vale write-good.Passive = NO -->
 
diff --git a/hive-c0re/src/stats/otel_metrics.rs b/hive-c0re/src/stats/otel_metrics.rs
index 36ae37bb..40c4d6bd 100644
--- a/hive-c0re/src/stats/otel_metrics.rs
+++ b/hive-c0re/src/stats/otel_metrics.rs
@@ -347,8 +347,9 @@ fn attrs(c: &ContainerResource) -> Vec<KeyValue> {
 
 /// Resource: `service.name = hyperhive-c0re`, the host architecture, the
 /// running hyperhive revision, plus whatever the operator set in
-/// `extraResourceAttributes` (where the hive `hive` / `swarm` labels live, same
-/// channel the agent-side export uses).
+/// `swarm.hiveOtel.extraResourceAttributes` (swarm-wide labels, same channel
+/// the agent-side export uses; the `hive` label is stamped by the swarm
+/// collector, not set here).
 ///
 /// These belong on the resource rather than on each data point for the same
 /// reason `hive` / `swarm` do: they are constant across one `hive-c0re`. Only