diff --git a/captain_hook/grants/judge.py b/captain_hook/grants/judge.py
index 362e6850..9f6efe91 100644
--- a/captain_hook/grants/judge.py
+++ b/captain_hook/grants/judge.py
@@ -95,7 +95,7 @@ class Judge:
     rules: str
     contexts: Sequence[PromptContext] = ()
     model: TModel = "small"
-    specialty: TSpecialty = "general"
+    specialty: TSpecialty = "review"
     deadline: float = 20
     transcript: bool | int | Literal["recent", "full"] = False
     root_transcript: bool | int | Literal["recent", "full"] = False
diff --git a/tests/test_grants.py b/tests/test_grants.py
index 45115743..5a7cf733 100644
--- a/tests/test_grants.py
+++ b/tests/test_grants.py
@@ -524,3 +524,10 @@ def test_linking_keeps_a_concurrent_revocation(tmp_path: Path) -> None:
     store.revoke(grant.id)
     linked = store.link(grant.id, "cc-slack", "daemon-1")
     assert linked.revoked is not None and linked.links == {"cc-slack": "daemon-1"}
+
+
+def test_the_judge_runs_on_luna_at_low_effort() -> None:
+    from spawnllm import LlmBackends
+
+    judge = Judge("rules")
+    assert LlmBackends.for_specialty(judge.specialty).resolve_model(judge.model) == "gpt-6-luna:low"