fix: reset stale remote agent updates after sac-api restart (0.4.3)

Clear zombie running jobs on deploy and add admin API to cancel a stuck job.

Co-authored-by: Cursor <cursoragent@cursor.com>

backend/app/api/v1/hosts.py

@@ -22,6 +22,8 @@ from app.services.agent_update import (
 )
 from app.services.host_remote_actions import (
     RemoteActionAlreadyRunningError,
+    cancel_host_remote_action,
+    clear_stale_running_remote_actions,
     get_remote_action_status,
     run_agent_fallback_action,
     run_ssh_monitor_update_action,
@@ -605,6 +607,21 @@ def get_host_remote_job(
     return HostRemoteActionJobResponse(**get_remote_action_status(host))
 
 
+@router.post("/{host_id}/actions/remote-job/cancel", response_model=HostRemoteActionJobResponse)
+def cancel_host_remote_job(
+    host_id: int,
+    db: Session = Depends(get_db),
+    _user=Depends(require_admin),
+) -> HostRemoteActionJobResponse:
+    host = db.get(Host, host_id)
+    if host is None:
+        raise HTTPException(status_code=404, detail="Host not found")
+    if not cancel_host_remote_action(db, host):
+        raise HTTPException(status_code=409, detail="No running remote action for this host")
+    db.refresh(host)
+    return HostRemoteActionJobResponse(**get_remote_action_status(host))
+
+
 @router.post("/{host_id}/actions/sessions/list", response_model=HostSessionsResponse)
 def list_host_sessions(
     host_id: int,