From 2b07b441274f6bfb772e949b981702aba0d9ba2e Mon Sep 17 00:00:00 2001 From: Hitesh Date: Mon, 21 Sep 2026 13:29:16 +0530 Subject: [PATCH] fix(shadow-eval): preserve chat bridge tool payloads --- litellm/integrations/shadow_eval_logger.py | 2 + .../integrations/test_shadow_eval_logger.py | 48 +++++++++++++++++++ 2 files changed, 50 insertions(+) diff --git a/litellm/integrations/shadow_eval_logger.py b/litellm/integrations/shadow_eval_logger.py index cdc108a6b4e..67e67d3e306 100644 --- a/litellm/integrations/shadow_eval_logger.py +++ b/litellm/integrations/shadow_eval_logger.py @@ -149,6 +149,8 @@ def _chat_request_from_responses( from litellm.types.llms.openai import ResponsesAPIOptionalRequestParams wire_body: Final = _proxy_wire_body(kwargs) + if "messages" in wire_body: + return MappingProxyType({**wire_body, "messages": _chat_messages(kwargs)}) instructions: Final = kwargs.get("instructions") or wire_body.get("instructions") responses_request: Final = MappingProxyType( dict( diff --git a/tests/test_litellm/integrations/test_shadow_eval_logger.py b/tests/test_litellm/integrations/test_shadow_eval_logger.py index 76efe9c8576..9a14ef8288d 100644 --- a/tests/test_litellm/integrations/test_shadow_eval_logger.py +++ b/tests/test_litellm/integrations/test_shadow_eval_logger.py @@ -386,6 +386,54 @@ class TestSurfaceNormalization: assert "stream" not in shadow_call assert shadow_call["metadata"][INTERNAL_CALL_ORIGIN_METADATA_KEY] == SHADOW_EVAL_ROUTER_CALL_ORIGIN + async def test_responses_arm_preserves_chat_tool_schema_from_bridge_wire_body(self): + from litellm.types.llms.openai import ResponsesAPIResponse + + hook_kwargs = _success_kwargs(call_type="aresponses") + hook_kwargs["messages"] = [ + {"role": "user", "content": "look up"}, + { + "role": "assistant", + "tool_calls": [ + {"id": "call_123", "type": "function", "function": {"name": "lookup", "arguments": "{}"}} + ], + }, + {"role": "tool", "tool_call_id": "call_123", "content": "42"}, + ] + hook_kwargs["litellm_params"]["proxy_server_request"] = { + "body": { + "model": "gpt-5.6-sol", + "messages": hook_kwargs["messages"], + "max_completion_tokens": 128, + "tool_choice": {"type": "function", "function": {"name": "lookup"}}, + "tools": [{"type": "function", "function": {"name": "lookup", "parameters": {}}}], + } + } + response = ResponsesAPIResponse.model_validate( + { + **RESPONSES_API_RESPONSE, + "output": [ + { + "type": "message", + "id": "msg_1", + "status": "completed", + "role": "assistant", + "content": [{"type": "output_text", "text": "42", "annotations": []}], + } + ], + } + ) + + _, router = await self._drive(hook_kwargs, response) + + shadow_call = router.acompletion.call_args_list[0].kwargs + assert shadow_call["max_completion_tokens"] == 128 + assert shadow_call["tool_choice"] == {"type": "function", "function": {"name": "lookup"}} + assert shadow_call["tools"] == hook_kwargs["litellm_params"]["proxy_server_request"]["body"]["tools"] + assert shadow_call["tools"][0]["function"]["name"] == "lookup" + assert shadow_call["messages"][1]["tool_calls"][0]["id"] == "call_123" + assert shadow_call["messages"][2]["tool_call_id"] == "call_123" + async def test_responses_arm_translates_wire_body_params_and_drops_surface_only_keys(self): from litellm.types.llms.openai import ResponsesAPIResponse