fix(responses-bridge): emit annotations on output_text input items

Closes #35213

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm/completion_extras/litellm_responses_transformation/transformation.py

@@ -740,7 +740,7 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
         if role == "user" or role == "system" or role == "tool":
             return {"type": "input_text", "text": content}
         else:
-            return {"type": "output_text", "text": content}
+            return {"type": "output_text", "text": content, "annotations": []}
 
     def _convert_content_to_responses_format_image(
         self, content: "ChatCompletionImageObject", role: str
@@ -851,9 +851,9 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
                             "computer_screenshot",
                             "summary_text",
                         ]:
-                            # Already in responses API format
-                            result.append(item)
-                            verbose_logger.debug(f"Chat provider:   passthrough -> {item}")
+                            passthrough = {"annotations": [], **item} if item_type == "output_text" else item
+                            result.append(passthrough)
+                            verbose_logger.debug(f"Chat provider:   passthrough -> {passthrough}")
                         else:
                             # Default to input_text for unknown types
                             converted = self._convert_content_str_to_input_text(str(item.get("text", item)), role)

tests/test_litellm/completion_extras/litellm_responses_transformation/test_completion_extras_litellm_responses_transformation_transformation.py

@@ -2962,3 +2962,79 @@ async def test_acompletion_bridge_normalizes_stream_options_on_the_wire(
         assert "stream_options" not in request_body
     else:
         assert request_body["stream_options"] == expected_wire_stream_options
+
+
+def test_assistant_history_output_text_items_include_annotations():
+    """
+    Regression test for issue #35213: assistant history replayed through the
+    chat -> responses bridge produced ``output_text`` blocks without the
+    ``annotations`` key, which strict Responses backends reject with
+    "Required property 'annotations' is missing".
+    """
+    from openai.types.responses.response_output_text_param import ResponseOutputTextParam
+    from pydantic import TypeAdapter
+
+    from litellm.completion_extras.litellm_responses_transformation.transformation import (
+        LiteLLMResponsesTransformationHandler,
+    )
+
+    handler = LiteLLMResponsesTransformationHandler()
+
+    messages = [
+        {"role": "user", "content": "first question"},
+        {"role": "assistant", "content": "plain string answer"},
+        {"role": "user", "content": "second question"},
+        {"role": "assistant", "content": [{"type": "text", "text": "list text answer"}]},
+        {"role": "user", "content": "third question"},
+        {"role": "assistant", "content": [{"type": "output_text", "text": "passthrough answer"}]},
+    ]
+
+    input_items, _ = handler.convert_chat_completion_messages_to_responses_api(messages)
+
+    output_text_blocks = [
+        block
+        for item in input_items
+        if item.get("role") == "assistant"
+        for block in item["content"]
+        if block.get("type") == "output_text"
+    ]
+    assert len(output_text_blocks) == 3
+    assert [block["text"] for block in output_text_blocks] == [
+        "plain string answer",
+        "list text answer",
+        "passthrough answer",
+    ]
+    output_text_adapter = TypeAdapter(ResponseOutputTextParam)
+    for block in output_text_blocks:
+        assert block["annotations"] == []
+        output_text_adapter.validate_python(block)
+
+
+def test_assistant_history_output_text_preserves_existing_annotations():
+    from litellm.completion_extras.litellm_responses_transformation.transformation import (
+        LiteLLMResponsesTransformationHandler,
+    )
+
+    annotations = [
+        {
+            "type": "url_citation",
+            "url": "https://example.com",
+            "title": "example",
+            "start_index": 0,
+            "end_index": 5,
+        }
+    ]
+    handler = LiteLLMResponsesTransformationHandler()
+
+    input_items, _ = handler.convert_chat_completion_messages_to_responses_api(
+        [
+            {"role": "user", "content": "hi"},
+            {
+                "role": "assistant",
+                "content": [{"type": "output_text", "text": "cited", "annotations": annotations}],
+            },
+        ]
+    )
+
+    assistant_item = next(item for item in input_items if item.get("role") == "assistant")
+    assert assistant_item["content"][0]["annotations"] == annotations