From 3190f42abf65e25053e59c2e8b5c9a96dd21c220 Mon Sep 17 00:00:00 2001
From: mateo <mateo@berri.ai>
Date: Thu, 3 Sep 2026 08:09:51 +0000
Subject: [PATCH] refactor: clear fresh tech debt from the last 24 hours
 (2026-09-03)

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 .../model_armor/model_armor.py                |  5 +---
 litellm/responses/streaming_iterator.py       | 19 +++++-------
 litellm/rust_bridge/runtime.py                | 30 -------------------
 type-discipline-budget.json                   |  4 +--
 4 files changed, 10 insertions(+), 48 deletions(-)

diff --git a/litellm/proxy/guardrails/guardrail_hooks/model_armor/model_armor.py b/litellm/proxy/guardrails/guardrail_hooks/model_armor/model_armor.py
index 7d88a037f4f..f0de6ee0ac7 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/model_armor/model_armor.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/model_armor/model_armor.py
@@ -1095,10 +1095,7 @@ class ModelArmorGuardrail(CustomGuardrail, VertexBase):
             add_guardrail_to_applied_guardrails_header,
         )
 
-        # Collect all chunks
-        all_chunks: Final[list[Any]] = []
-        async for chunk in response:
-            all_chunks.append(chunk)
+        all_chunks: Final[Sequence[Any]] = tuple([chunk async for chunk in response])
 
         if not all_chunks or self._is_terminal_error_stream(all_chunks):
             for chunk in all_chunks:
diff --git a/litellm/responses/streaming_iterator.py b/litellm/responses/streaming_iterator.py
index f271655f5e3..9f9016c5a7f 100644
--- a/litellm/responses/streaming_iterator.py
+++ b/litellm/responses/streaming_iterator.py
@@ -496,10 +496,9 @@ class BaseResponsesAPIStreamingIterator:
         if logging_response is self.completed_response:
             return
         target: Final[object] = getattr(logging_response, "response", None)
-        existing_hidden: Final[object] = getattr(target, "_hidden_params", None)
-        if not isinstance(existing_hidden, Mapping):
+        if not isinstance(target, ResponsesAPIResponse):
             return
-        existing: Final[Mapping[str, object]] = existing_hidden
+        existing: Final[Mapping[str, object]] = target._hidden_params
         source_hidden: Final[object] = getattr(
             getattr(self.completed_response, "response", None), "_hidden_params", None
         )
@@ -510,15 +509,11 @@ class BaseResponsesAPIStreamingIterator:
         raw_headers: Final[Mapping[str, object]] = raw if isinstance(raw, Mapping) else EMPTY_MAPPING
         # rebuild by value and let existing keys win: sharing the source dicts would alias what the proxy
         # splats into the client's HTTP headers, and copying non-header keys would carry response_cost
-        setattr(  # noqa: B010  # target is typed object here, so a plain attribute store does not type check
-            target,
-            "_hidden_params",
-            {  # mutable-ok: the cost calculator writes optional_params into _hidden_params
-                "additional_headers": {**headers},  # mutable-ok: fresh copy, logging callbacks may mutate it
-                "headers": {**raw_headers},  # mutable-ok: fresh copy, logging callbacks may mutate it
-                **existing,
-            },
-        )
+        target._hidden_params = {  # mutable-ok: the cost calculator writes optional_params into _hidden_params
+            "additional_headers": {**headers},  # mutable-ok: fresh copy, logging callbacks may mutate it
+            "headers": {**raw_headers},  # mutable-ok: fresh copy, logging callbacks may mutate it
+            **existing,
+        }
 
     def _handle_logging_completed_response(self):
         """Base implementation - should be overridden by subclasses"""
diff --git a/litellm/rust_bridge/runtime.py b/litellm/rust_bridge/runtime.py
index 00f06c046a2..d411673439f 100644
--- a/litellm/rust_bridge/runtime.py
+++ b/litellm/rust_bridge/runtime.py
@@ -116,28 +116,6 @@ async def aattempt(
     return RustHandled(adapt(value))
 
 
-def call(operation: Callable[[], ResultT], context: BridgeErrorContext) -> ResultT:
-    exceptions: Final = native_exception_types()
-    if exceptions is None:
-        return operation()
-    upstream: Final = exceptions[1]
-    try:
-        return operation()
-    except upstream as error:
-        _raise_upstream(error, context)
-
-
-async def acall(operation: Callable[[], Awaitable[ResultT]], context: BridgeErrorContext) -> ResultT:
-    exceptions: Final = native_exception_types()
-    if exceptions is None:
-        return await operation()
-    upstream: Final = exceptions[1]
-    try:
-        return await operation()
-    except upstream as error:
-        _raise_upstream(error, context)
-
-
 def _decline_reason(error: BaseException) -> str:
     reason: Final[object] = error.args[0] if error.args else str(error)
     return reason if isinstance(reason, str) else str(reason)
@@ -170,11 +148,3 @@ def _raise_upstream(error: BaseException, context: BridgeErrorContext) -> NoRetu
         llm_provider=context.provider,
         model=context.model,
     ) from error
-
-
-def identity(value: ResultT) -> ResultT:
-    return value
-
-
-async def async_none() -> None:
-    return None
diff --git a/type-discipline-budget.json b/type-discipline-budget.json
index d5bf3883be4..704d7e8a596 100644
--- a/type-discipline-budget.json
+++ b/type-discipline-budget.json
@@ -1,9 +1,9 @@
 {
   "LIT001": {
-    "limit": 22330
+    "limit": 22329
   },
   "LIT002": {
-    "limit": 26763
+    "limit": 26762
   },
   "LIT003": {
     "limit": 261