feat(agent): show and manage Swarm services scaled to 0 replicas

Zero-replica Swarm services now appear in the dashboard as "stopped"
instead of being invisible. Start/stop/restart/pull/logs all handle
the synthetic swarm: IDs correctly. Uses low-level API for container
listing to avoid NotFound errors from Dead task containers.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

agent/docker_ops.py

@@ -84,16 +84,31 @@ class DockerOps:
     def list_containers(self) -> list[dict]:
         """List all containers (running + stopped) with metadata.
 
+        Also includes Swarm services scaled to 0 replicas (which have no
+        containers) so they remain visible in the dashboard.
+
         Returns a list of dicts, each containing:
             id, name, status, image, created, uptime, is_swarm,
             swarm_service
+
+        Uses the low-level API to get container IDs to avoid NotFound errors
+        from containers that disappear between list and inspect (e.g. Swarm
+        task containers in the Dead state).
         """
-        containers = self.client.containers.list(all=True)
+        raw_ids = [r["Id"] for r in self.client.api.containers(all=True)]
         result = []
-        for c in containers:
+        swarm_services_with_containers = set()
+
+        for cid in raw_ids:
+            try:
+                c = self.client.containers.get(cid)
+            except NotFound:
+                continue
             image_tags = c.image.tags
             image_name = image_tags[0] if image_tags else "<none>"
             swarm_service = self._swarm_service_name(c)
+            if swarm_service:
+                swarm_services_with_containers.add(swarm_service)
             started_at = c.attrs.get("State", {}).get("StartedAt", "")
             result.append(
                 {
@@ -107,10 +122,53 @@ class DockerOps:
                     "swarm_service": swarm_service,
                 }
             )
+
+        # Include Swarm services with 0 replicas (only works on manager nodes)
+        try:
+            for svc in self.client.services.list():
+                if svc.name not in swarm_services_with_containers:
+                    image = (
+                        svc.attrs.get("Spec", {})
+                        .get("TaskTemplate", {})
+                        .get("ContainerSpec", {})
+                        .get("Image", "<none>")
+                    )
+                    if "@" in image:
+                        image = image.split("@")[0]
+                    result.append(
+                        {
+                            "id": f"swarm:{svc.name}",
+                            "name": svc.name,
+                            "status": "stopped",
+                            "image": image,
+                            "created": svc.attrs.get("CreatedAt", ""),
+                            "uptime": None,
+                            "is_swarm": True,
+                            "swarm_service": svc.name,
+                        }
+                    )
+        except Exception:
+            pass  # Not a manager node or Swarm not active
+
         return result
 
     def start_container(self, container_id: str) -> dict:
         """Start a container.  For Swarm services, scale to 1 replica."""
+        # Handle 0-replica Swarm services (synthetic swarm: IDs)
+        if container_id.startswith("swarm:"):
+            svc_name = container_id[6:]
+            try:
+                service = self._get_service(svc_name)
+                if service:
+                    service.scale(1)
+                    return {
+                        "success": True,
+                        "message": f"Scaled Swarm service {svc_name} to 1 replica",
+                    }
+                return {"success": False, "message": f"Swarm service {svc_name} not found"}
+            except Exception as exc:
+                return {"success": False, "message": str(exc)}
+
         try:
             container = self.client.containers.get(container_id)
         except NotFound:
@@ -135,6 +193,20 @@ class DockerOps:
 
     def stop_container(self, container_id: str) -> dict:
         """Stop a container.  For Swarm services, scale to 0 replicas."""
+        if container_id.startswith("swarm:"):
+            svc_name = container_id[6:]
+            try:
+                service = self._get_service(svc_name)
+                if service:
+                    service.scale(0)
+                    return {
+                        "success": True,
+                        "message": f"Scaled Swarm service {svc_name} to 0 replicas",
+                    }
+                return {"success": False, "message": f"Swarm service {svc_name} not found"}
+            except Exception as exc:
+                return {"success": False, "message": str(exc)}
+
         try:
             container = self.client.containers.get(container_id)
         except NotFound:
@@ -159,6 +231,20 @@ class DockerOps:
 
     def restart_container(self, container_id: str) -> dict:
         """Restart a container.  For Swarm services, force-update."""
+        if container_id.startswith("swarm:"):
+            svc_name = container_id[6:]
+            try:
+                service = self._get_service(svc_name)
+                if service:
+                    service.force_update()
+                    return {
+                        "success": True,
+                        "message": f"Force-updated Swarm service {svc_name}",
+                    }
+                return {"success": False, "message": f"Swarm service {svc_name} not found"}
+            except Exception as exc:
+                return {"success": False, "message": str(exc)}
+
         try:
             container = self.client.containers.get(container_id)
         except NotFound:
@@ -187,6 +273,13 @@ class DockerOps:
         Returns:
             dict with keys ``container`` (name) and ``logs`` (string).
         """
+        if container_id.startswith("swarm:"):
+            svc_name = container_id[6:]
+            return {
+                "container": svc_name,
+                "logs": "Service is scaled to 0 replicas — no logs available.\nStart the service to view logs.",
+            }
+
         try:
             container = self.client.containers.get(container_id)
         except NotFound:
@@ -200,6 +293,27 @@ class DockerOps:
 
     def pull_image(self, container_id: str) -> dict:
         """Pull the latest version of the container's image."""
+        if container_id.startswith("swarm:"):
+            svc_name = container_id[6:]
+            service = self._get_service(svc_name)
+            if not service:
+                return {"success": False, "message": f"Swarm service {svc_name} not found"}
+            image = (
+                service.attrs.get("Spec", {})
+                .get("TaskTemplate", {})
+                .get("ContainerSpec", {})
+                .get("Image", "")
+            )
+            if "@" in image:
+                image = image.split("@")[0]
+            if not image:
+                return {"success": False, "message": f"No image found for service {svc_name}"}
+            try:
+                self.client.images.pull(image)
+                return {"success": True, "message": f"Pulled image {image}"}
+            except Exception as exc:
+                return {"success": False, "message": str(exc)}
+
         try:
             container = self.client.containers.get(container_id)
         except NotFound: