fix(mypy): fix remaining type errors after first pass

- Perplexity: avoid TypedDict spread by using dict() conversion
- Vertex batch_embed: use Any type for request_data variable
- route_llm_request: sync route_request Literal with base_process_llm_request
- Presidio: cast chunks from internal generators to ModelResponseStream
- key_management: properly handle None case for object_permission_dict
- completion_transformation: use isinstance check for list type narrowing

Co-authored-by: yuneng-jiang <yuneng-jiang@users.noreply.github.com>

litellm/llms/perplexity/responses/transformation.py

@@ -71,8 +71,11 @@ class PerplexityResponsesConfig(OpenAIResponsesAPIConfig):
             result: List[Any] = []
             for item in input:
                 if isinstance(item, dict) and "type" not in item:
-                    item = {**item, "type": "message"}
-                result.append(item)
+                    new_item = dict(item)  # convert to plain dict to avoid TypedDict checking
+                    new_item["type"] = "message"
+                    result.append(new_item)
+                else:
+                    result.append(item)
             return result
         return input
 

litellm/llms/vertex_ai/gemini_embeddings/batch_embed_content_handler.py

@@ -201,7 +201,7 @@ class GoogleBatchEmbeddings(VertexLLM):
             )
 
         ### TRANSFORMATION (sync path) ###
-        request_data: Dict[str, Any]
+        request_data: Any
         if use_embed_content:
             resolved_files = {}
             if api_key:

litellm/proxy/guardrails/guardrail_hooks/presidio.py

@@ -1242,7 +1242,7 @@ class _OPTIONAL_PresidioPIIMasking(CustomGuardrail):
             async for chunk in self._stream_apply_output_masking(
                 response, request_data
             ):
-                yield chunk
+                yield cast(ModelResponseStream, chunk)
             return
 
         metadata = (request_data.get("metadata") or {}) if request_data else {}
@@ -1257,7 +1257,7 @@ class _OPTIONAL_PresidioPIIMasking(CustomGuardrail):
             return
 
         async for chunk in self._stream_pii_unmasking(response, request_data):
-            yield chunk
+            yield cast(ModelResponseStream, chunk)
 
     @staticmethod
     def _preserve_usage_from_last_chunk(

litellm/proxy/management_endpoints/key_management_endpoints.py

@@ -1776,11 +1776,13 @@ async def _validate_mcp_servers_for_key_update(
             user_api_key_cache=user_api_key_cache,
             check_db_only=True,
         )
-    object_permission_dict = (
-        data.object_permission.model_dump()
-        if data.object_permission is not None and hasattr(data.object_permission, "model_dump")
-        else data.object_permission
-    )
+    object_permission_dict: Optional[dict] = None
+    if data.object_permission is not None:
+        object_permission_dict = (
+            data.object_permission.model_dump()
+            if hasattr(data.object_permission, "model_dump")
+            else dict(data.object_permission)  # type: ignore[arg-type]
+        )
     await validate_key_mcp_servers_against_team(
         object_permission=object_permission_dict,
         team_obj=effective_team_obj,

litellm/proxy/route_llm_request.py

@@ -173,8 +173,21 @@ async def route_request(  # noqa: PLR0915 - Complex routing function, refactorin
         "agenerate_content",
         "agenerate_content_stream",
         "allm_passthrough_route",
+        "acreate_batch",
+        "aretrieve_batch",
+        "alist_batches",
+        "afile_content",
+        "afile_retrieve",
+        "acreate_fine_tuning_job",
+        "acancel_fine_tuning_job",
+        "alist_fine_tuning_jobs",
+        "aretrieve_fine_tuning_job",
         "avector_store_search",
         "avector_store_create",
+        "avector_store_retrieve",
+        "avector_store_list",
+        "avector_store_update",
+        "avector_store_delete",
         "avector_store_file_create",
         "avector_store_file_list",
         "avector_store_file_retrieve",
@@ -207,6 +220,8 @@ async def route_request(  # noqa: PLR0915 - Complex routing function, refactorin
         "aget_interaction",
         "adelete_interaction",
         "acancel_interaction",
+        "asend_message",
+        "call_mcp_tool",
         "acancel_batch",
         "afile_delete",
         "acreate_eval",

litellm/responses/litellm_completion_transformation/transformation.py

@@ -410,11 +410,12 @@ class LiteLLMCompletionResponsesConfig:
                                     else getattr(new_msg, "role", None)
                                 )
                                 if new_role == "assistant":
-                                    new_tcs: list = (
+                                    _raw_tcs = (
                                         new_msg.get("tool_calls")
                                         if isinstance(new_msg, dict)
                                         else getattr(new_msg, "tool_calls", None)
-                                    ) or []
+                                    )
+                                    new_tcs: list = _raw_tcs if isinstance(_raw_tcs, list) else []
                                     for tc in new_tcs:
                                         LiteLLMCompletionResponsesConfig._add_tool_call_to_assistant(
                                             last_msg, tc