diff --git a/app/api/search.py b/app/api/search.py
index b781986..023ffc9 100644
--- a/app/api/search.py
+++ b/app/api/search.py
@@ -647,6 +647,39 @@ async def ask(
       - 일반 client 의 X-Source=eval 시도는 무시되고 source='document_server' 로 강제.
       - source != 'eval' 이면 eval_case_id 항상 None.
     """
+    # 오케스트레이션은 _run_ask 로 분리(라우터=입력+deps 해소만). 동작 무변경 (refactor: ask-orchestrator).
+    return await _run_ask(
+        q=q,
+        user=user,
+        session=session,
+        background_tasks=background_tasks,
+        limit=limit,
+        debug=debug,
+        backend=backend,
+        corpus_variant=corpus_variant,
+        exact_knn=exact_knn,
+        x_source=x_source,
+        x_eval_case_id=x_eval_case_id,
+        x_eval_token=x_eval_token,
+    )
+
+
+async def _run_ask(
+    q: str,
+    user: User,
+    session: AsyncSession,
+    background_tasks: BackgroundTasks,
+    limit: int,
+    debug: bool,
+    backend: str | None,
+    corpus_variant: str | None,
+    exact_knn: bool,
+    x_source: str | None,
+    x_eval_case_id: str | None,
+    x_eval_token: str | None,
+) -> "AskResponse | JSONResponse":
+    """/ask 오케스트레이션 — 검색→evidence/classifier→refusal→synthesis→grounding/verifier→
+    7-tier 재게이트→telemetry. ask() 가 FastAPI deps 해소 후 호출. 동작은 기존 핸들러와 동일."""
     t_total = time.perf_counter()
     defense_log: dict = {}  # per-layer flag snapshot
     source, eval_case_id = _resolve_eval_identity(x_source, x_eval_case_id, x_eval_token)