From 4064e59aa1c18099c53866880ff8ff5f4e9a1d0c Mon Sep 17 00:00:00 2001
From: Chau Vu / CPF-FAMILY <58875435+chauvuusvn@users.noreply.github.com>
Date: Thu, 8 Oct 2026 21:47:06 +0700
Subject: [PATCH] fix(proxy): add /openai/responses to API_ROUTE_TO_CALL_TYPES
 and Prometheus route normalization

- Map /openai/responses routes in API_ROUTE_TO_CALL_TYPES for tool allowlist and guardrail validation
- Add regex normalization patterns for /openai/responses/{response_id} in normalize_request_route
- Expand test suite in tests/proxy_unit_tests/test_openai_responses_routes.py
---
 litellm/proxy/auth/auth_utils.py                 |  3 +++
 litellm/types/utils.py                           |  3 +++
 .../test_openai_responses_routes.py              | 16 ++++++++++++++++
 3 files changed, 22 insertions(+)

diff --git a/litellm/proxy/auth/auth_utils.py b/litellm/proxy/auth/auth_utils.py
index 2bd10b90101..9e603af01dd 100644
--- a/litellm/proxy/auth/auth_utils.py
+++ b/litellm/proxy/auth/auth_utils.py
@@ -857,6 +857,9 @@ def normalize_request_route(route: str) -> str:
         (r"^(/(?:openai/)?v1/responses)/([^/]+)(/input_items)$", r"\1/{response_id}\3"),
         (r"^(/(?:openai/)?v1/responses)/([^/]+)(/cancel)$", r"\1/{response_id}\3"),
         (r"^(/(?:openai/)?v1/responses)/([^/]+)$", r"\1/{response_id}"),
+        (r"^(/openai/responses)/([^/]+)(/input_items)$", r"\1/{response_id}\3"),
+        (r"^(/openai/responses)/([^/]+)(/cancel)$", r"\1/{response_id}\3"),
+        (r"^(/openai/responses)/([^/]+)$", r"\1/{response_id}"),
         (r"^(/responses)/([^/]+)(/input_items)$", r"\1/{response_id}\3"),
         (r"^(/responses)/([^/]+)(/cancel)$", r"\1/{response_id}\3"),
         (r"^(/responses)/([^/]+)$", r"\1/{response_id}"),
diff --git a/litellm/types/utils.py b/litellm/types/utils.py
index a99c22e9479..7ebbe04b197 100644
--- a/litellm/types/utils.py
+++ b/litellm/types/utils.py
@@ -1038,12 +1038,15 @@ API_ROUTE_TO_CALL_TYPES: Final[Mapping[str, Sequence[CallTypes]]] = {
     "/responses": (CallTypes.aresponses, CallTypes.responses),
     "/v1/responses": (CallTypes.aresponses, CallTypes.responses),
     "/openai/v1/responses": (CallTypes.aresponses, CallTypes.responses),
+    "/openai/responses": (CallTypes.aresponses, CallTypes.responses),
     "/responses/{response_id}": (CallTypes.aresponses, CallTypes.responses),
     "/v1/responses/{response_id}": (CallTypes.aresponses, CallTypes.responses),
     "/openai/v1/responses/{response_id}": (CallTypes.aresponses, CallTypes.responses),
+    "/openai/responses/{response_id}": (CallTypes.aresponses, CallTypes.responses),
     "/responses/{response_id}/input_items": (CallTypes.alist_input_items,),
     "/v1/responses/{response_id}/input_items": (CallTypes.alist_input_items,),
     "/openai/v1/responses/{response_id}/input_items": (CallTypes.alist_input_items,),
+    "/openai/responses/{response_id}/input_items": (CallTypes.alist_input_items,),
     # Realtime API
     "/realtime": [CallTypes.arealtime],
     "/v1/realtime": [CallTypes.arealtime],
diff --git a/tests/proxy_unit_tests/test_openai_responses_routes.py b/tests/proxy_unit_tests/test_openai_responses_routes.py
index dc83a1fb257..21c7865a7fb 100644
--- a/tests/proxy_unit_tests/test_openai_responses_routes.py
+++ b/tests/proxy_unit_tests/test_openai_responses_routes.py
@@ -1,8 +1,10 @@
 from litellm.proxy._types import LiteLLMRoutes
 from litellm.proxy.agent_endpoints.auth.managed_authorization import _MANAGED_MODEL_ROUTES
+from litellm.proxy.auth.auth_utils import normalize_request_route
 from litellm.proxy.common_utils.custom_openapi_spec import CustomOpenAPISpec
 from litellm.proxy.response_api_endpoints.endpoints import router
 from litellm.proxy.spend_tracking.budget_reservation import _UNBILLED_ROUTES
+from litellm.types.utils import API_ROUTE_TO_CALL_TYPES, CallTypes
 
 
 def test_openai_responses_routes_registered():
@@ -25,3 +27,17 @@ def test_openai_responses_routes_registered():
     assert "/openai/responses" in _MANAGED_MODEL_ROUTES
     assert "/openai/responses/input_tokens" in _UNBILLED_ROUTES
     assert "/openai/responses" in LiteLLMRoutes.openai_routes.value
+
+
+def test_openai_responses_api_route_to_call_types():
+    """Verify that /openai/responses mappings exist in API_ROUTE_TO_CALL_TYPES for guardrails/tools."""
+    assert API_ROUTE_TO_CALL_TYPES.get("/openai/responses") == (CallTypes.aresponses, CallTypes.responses)
+    assert API_ROUTE_TO_CALL_TYPES.get("/openai/responses/{response_id}") == (CallTypes.aresponses, CallTypes.responses)
+    assert API_ROUTE_TO_CALL_TYPES.get("/openai/responses/{response_id}/input_items") == (CallTypes.alist_input_items,)
+
+
+def test_openai_responses_metrics_normalization():
+    """Verify that dynamic response IDs are normalized to prevent Prometheus high-cardinality metrics."""
+    assert normalize_request_route("/openai/responses/resp-123456") == "/openai/responses/{response_id}"
+    assert normalize_request_route("/openai/responses/resp-123456/cancel") == "/openai/responses/{response_id}/cancel"
+    assert normalize_request_route("/openai/responses/resp-123456/input_items") == "/openai/responses/{response_id}/input_items"