diff --git a/tests/control_plane/test_todo_claim_argument_recovery.py b/tests/control_plane/test_todo_claim_argument_recovery.py
index 08d0f8125c..fe974441f1 100644
--- a/tests/control_plane/test_todo_claim_argument_recovery.py
+++ b/tests/control_plane/test_todo_claim_argument_recovery.py
@@ -210,3 +210,22 @@ def test_ordinary_claim_progress_completion_need_no_lease_flags(claim_cli):
     replay = run([*prefix, "todo", "complete", *common, "--evidence", "evidence:validated-output"])
     assert replay["ok"] is True
     assert snapshot() == before
+
+
+@pytest.mark.parametrize("command", ["update", "complete"])
+def test_empty_managed_reference_is_rejected_without_mutation(claim_cli, command):
+    _, run, snapshot, prefix, claim, _, _, _ = claim_cli
+    accepted = run([*prefix, *claim, "--claimed-by", "peer-a",
+        "--claim-operation-id", "valid-managed-reference"])
+    assert accepted["ok"] and accepted["execution_managed"]
+    before = snapshot()
+    arguments = [*prefix, "todo", command, "--goal-id", GOAL, "--todo-id", TODO,
+        "--agent-id", "peer-a", "--claim-operation-id", ""]
+    arguments += (["--note", "Only a valid original execution may update this task."]
+        if command == "update" else ["--evidence", "evidence:validated-output"])
+    rejected = run(arguments, expected_exit=1)
+    assert rejected["error_code"] == (
+        "invalid_local_coordination_todo_update_request" if command == "update"
+        else "invalid_local_coordination_todo_terminal_lifecycle_request")
+    assert rejected["error"] == "claim_operation_id must be a non-empty trimmed string"
+    assert snapshot() == before
diff --git a/tests/test_cli_argument_diagnostics.py b/tests/test_cli_argument_diagnostics.py
index c3b06e4847..3991d4b80d 100644
--- a/tests/test_cli_argument_diagnostics.py
+++ b/tests/test_cli_argument_diagnostics.py
@@ -557,7 +557,7 @@ def test_todo_claim_validation_preserves_exact_diagnostics(
     assert str(exc_info.value) == expected
 
 
-@pytest.mark.parametrize("command", ["add", "list", "update", "complete", "supersede"])
+@pytest.mark.parametrize("command", ["add", "list", "supersede"])
 @pytest.mark.parametrize("operation_id", ["retry-one", ""])
 def test_claim_operation_id_is_not_silently_ignored_by_other_commands(command, operation_id):
     args = build_parser().parse_args([
@@ -567,6 +567,18 @@ def test_claim_operation_id_is_not_silently_ignored_by_other_commands(command, o
         validate_shared_todo_options(args)
 
 
+@pytest.mark.parametrize("command", ["update", "complete"])
+@pytest.mark.parametrize("operation_id", ["retry-one", ""])
+def test_managed_claim_reference_is_retained_for_the_execution_owner(command, operation_id):
+    args = build_parser().parse_args([
+        "todo", command, "--goal-id", "example-goal", "--claim-operation-id", operation_id,
+    ])
+    validate_shared_todo_options(args)
+    # Scope validation admits the supported caller without dropping its raw
+    # reference. The native owner still rejects an invalid empty value.
+    assert args.claim_operation_id == operation_id
+
+
 @pytest.mark.parametrize(
     ("extra_args", "expected"),
     [