diff --git a/pioreactor/actions/leader/experiment_profile.py b/pioreactor/actions/leader/experiment_profile.py
index 725248f3..2b2540f1 100644
--- a/pioreactor/actions/leader/experiment_profile.py
+++ b/pioreactor/actions/leader/experiment_profile.py
@@ -892,8 +892,8 @@ def execute_experiment_profile(profile_filename: str, experiment: str, dry_run:
                 # stop all jobs started
                 # we can use active workers in experiment, since if a worker leaves an experiment or goes inactive, it's jobs are stopped
                 workers = get_active_workers_in_experiment(experiment)
-                with ClusterJobManager(workers) as jm:
-                    jm.kill_jobs(experiment=experiment, job_source="experiment_profile")
+                with ClusterJobManager() as cjm:
+                    cjm.kill_jobs(workers, experiment=experiment, job_source="experiment_profile")
 
             else:
                 if dry_run:
diff --git a/pioreactor/cli/pios.py b/pioreactor/cli/pios.py
index eb2485ba..1c663bc9 100644
--- a/pioreactor/cli/pios.py
+++ b/pioreactor/cli/pios.py
@@ -667,8 +667,10 @@ def kill(
             if confirm != "Y":
                 raise click.Abort()
 
-        with ClusterJobManager(units) as cm:
-            results = cm.kill_jobs(all_jobs=all_jobs, experiment=experiment, job_source=job_source, name=name)
+        with ClusterJobManager() as cm:
+            results = cm.kill_jobs(
+                units, all_jobs=all_jobs, experiment=experiment, job_source=job_source, name=name
+            )
 
         if json:
             for success, api_result in results:
diff --git a/pioreactor/tests/test_utils.py b/pioreactor/tests/test_utils.py
index 59612ee8..dd532007 100644
--- a/pioreactor/tests/test_utils.py
+++ b/pioreactor/tests/test_utils.py
@@ -269,8 +269,8 @@ def collect(msg):
 def test_ClusterJobManager_sends_requests():
     workers = ["pio01", "pio02", "pio03"]
     with capture_requests() as bucket:
-        with ClusterJobManager(workers) as cm:
-            cm.kill_jobs(name="stirring")
+        with ClusterJobManager() as cm:
+            cm.kill_jobs(workers, name="stirring")
 
     assert len(bucket) == len(workers)
     assert bucket[0].body is None
@@ -283,7 +283,7 @@ def test_ClusterJobManager_sends_requests():
 def test_empty_ClusterJobManager():
     workers = []
     with capture_requests() as bucket:
-        with ClusterJobManager(workers) as cm:
-            cm.kill_jobs(name="stirring")
+        with ClusterJobManager() as cm:
+            cm.kill_jobs(workers, name="stirring")
 
     assert len(bucket) == len(workers)
diff --git a/pioreactor/utils/__init__.py b/pioreactor/utils/__init__.py
index f5275295..c6e4b787 100644
--- a/pioreactor/utils/__init__.py
+++ b/pioreactor/utils/__init__.py
@@ -646,20 +646,20 @@ def __exit__(self, *args) -> None:
 
 
 class ClusterJobManager:
-    def __init__(self, units: tuple[str, ...]) -> None:
+    # this is a context manager to mimic the API for JobManager.
+    def __init__(self) -> None:
         if not whoami.am_I_leader():
             raise RoleError("Must be leader to use this. Maybe you want JobManager?")
 
-        self.units = units
-
+    @staticmethod
     def kill_jobs(
-        self,
+        units: tuple[str, ...],
         all_jobs: bool = False,
         experiment: str | None = None,
         name: str | None = None,
         job_source: str | None = None,
     ) -> list[tuple[bool, dict]]:
-        if len(self.units) == 0:
+        if len(units) == 0:
             return []
 
         if experiment:
@@ -680,8 +680,8 @@ def _thread_function(unit: str) -> tuple[bool, dict]:
                 print(f"Failed to send kill command to {unit}: {e}")
                 return False, {"unit": unit}
 
-        with ThreadPoolExecutor(max_workers=len(self.units)) as executor:
-            results = executor.map(_thread_function, self.units)
+        with ThreadPoolExecutor(max_workers=len(units)) as executor:
+            results = executor.map(_thread_function, units)
 
         return list(results)