diff --git a/cmd/kubectl-ate/README.md b/cmd/kubectl-ate/README.md
index 1e2842428..fa73cce1b 100644
--- a/cmd/kubectl-ate/README.md
+++ b/cmd/kubectl-ate/README.md
@@ -54,7 +54,7 @@ The kind overlay installed by `hack/install-ate-kind.sh --deploy-ate-system` alr
 Port-forward the Jaeger UI and invoke any command with `--trace`:
 ```bash
 kubectl port-forward -n otel-system svc/jaeger 16686:16686 &
-kubectl ate get actor my-counter-1 --trace
+kubectl ate get actor my-counter-1 -a demo --trace
 # open http://localhost:16686 and search for the most recent trace
 ```
 
@@ -108,7 +108,7 @@ kubectl ate get workers -l <label-selector>
 | `ATESPACE` | The atespace the actor belongs to. Part of the actor's identity; folded into the storage key as `actor:<atespace>:<name>`. |
 | `NAME` | The actor's name. User-provided for application actors; UUID for the golden actor that each template materialises during `ResumeGoldenActor`. |
 | `TEMPLATE` | The `ActorTemplate` the actor was created from, as `<namespace>/<name>` (the template namespace is distinct from `ATESPACE`). |
-| `STATUS` | One of `STATUS_RESUMING`, `STATUS_RUNNING`, `STATUS_SUSPENDING`, `STATUS_SUSPENDED`. |
+| `STATUS` | One of `STATUS_RESUMING`, `STATUS_RUNNING`, `STATUS_SUSPENDING`, `STATUS_SUSPENDED`, `STATUS_PAUSING`, `STATUS_PAUSED`, `STATUS_CRASHED`, `STATUS_DELETING`. |
 | `ATEOM POD` | The worker pod (namespace/name) currently hosting the actor. Empty while suspended. |
 | `ATEOM IP` | The pod IP of that worker. Empty while suspended. |
 | `VERSION` | Monotonic integer that increments on every state transition (resume / suspend / checkpoint). Useful for distinguishing snapshots. |
@@ -167,10 +167,20 @@ kubectl ate resume actor my-actor -a <atespace>
 # Suspend an actor (snapshots its state to storage and frees the worker)
 kubectl ate suspend actor my-actor -a <atespace>
 
+# Pause an actor (checkpoints it, but keeps the snapshot on the node VM)
+kubectl ate pause actor my-actor -a <atespace>
+
 # Delete an actor.
 kubectl ate delete actor my-actor -a <atespace>
 ```
 
+> **Note:** `suspend` and `pause` are both checkpoints, but they differ in where
+> the snapshot lands. `suspend` uploads it to snapshot storage, so the actor can
+> resume on any node. `pause` keeps it on the node VM, and the next `resume` is
+> prioritized onto that same node — cheaper and faster, but tied to one node's
+> availability. See [Lifecycle](../../docs/glossary.md#lifecycle) for the full
+> definitions.
+
 ### Actor Snapshots
 
 Suspending an actor creates a durable snapshot. Tags give snapshots stable,
@@ -208,6 +218,36 @@ kubectl ate logs actors my-actor -a <atespace> -f
 
 Logs are streamable only while the actor is bound to a worker (i.e., `STATUS_RUNNING`). For history across worker migrations, route through a centralized log backend (Cloud Logging, Loki, etc.); see `docs/observability.md`.
 
+### Resource Usage
+
+`kubectl ate top` reports CPU and memory usage of worker pods. `workers` is the
+only supported resource type.
+
+```bash
+# Usage for every worker pod
+kubectl ate top workers
+
+# Same filters as `get workers`: Kubernetes namespace, the atespace of the
+# assigned actor, or worker pool labels
+kubectl ate top workers -n <namespace>
+kubectl ate top workers -a <atespace>
+kubectl ate top workers -l <label-selector>
+```
+
+Metrics come from the Kubernetes metrics API, so the cluster needs
+metrics-server (or an equivalent) installed.
+
+#### `kubectl ate top workers` output columns
+
+| Column | Meaning |
+|---|---|
+| `NAME` | The worker pod name. |
+| `POOL` | The `WorkerPool` the pod belongs to. |
+| `STATUS` | `FREE` (idle, ready to receive an actor) or `ASSIGNED` (currently hosting an actor). |
+| `ASSIGNED ACTOR` | If `STATUS=ASSIGNED`, the actor reference `<namespace>/<template>/<actor-name>`. |
+| `CPU(CORES)` | CPU currently consumed by the worker pod. |
+| `MEMORY(bytes)` | Memory currently consumed by the worker pod. |
+
 ### Administration & Setup
 Commands for bootstrapping the Substrate control plane and debugging local environments.
 
diff --git a/demos/parking/README.md b/demos/parking/README.md
index 1942cfaa3..dd837c66a 100644
--- a/demos/parking/README.md
+++ b/demos/parking/README.md
@@ -71,8 +71,8 @@ Fill both workers by requesting two actors, leaving them `RUNNING`:
 curl -s -H "Host: p1.parking.actors.resources.substrate.ate.dev" http://localhost:8000
 curl -s -H "Host: p2.parking.actors.resources.substrate.ate.dev" http://localhost:8000
 
-kubectl ate get workers   # both workers are now bound to p1 and p2
-kubectl ate get actors    # p1,p2 RUNNING; p3,p4 SUSPENDED
+kubectl ate get workers             # both workers are now bound to p1 and p2
+kubectl ate get actors -a parking   # p1,p2 RUNNING; p3,p4 SUSPENDED
 ```
 
 Now request **p3** with timing. The pool is full, so this request **parks** —
diff --git a/hack/update/proto-fmt.sh b/hack/update/proto-fmt.sh
index 493c71d97..b30012035 100755
--- a/hack/update/proto-fmt.sh
+++ b/hack/update/proto-fmt.sh
@@ -44,7 +44,7 @@ files=(
 
 # Run clang-format on the found files.
 #
-# Don't reflow long lines, wince those tend to be comments which then require
+# Don't reflow long lines, since those tend to be comments which then require
 # re-running the proto generators, which makes "update-all" awkward.
 "${clangfmt}" \
     -i \