From aad5df50a066a26ed34a04f9e0076f997f43f6b5 Mon Sep 17 00:00:00 2001
From: Cursor Agent <cursoragent@cursor.com>
Date: Wed, 27 May 2026 15:10:26 +0000
Subject: [PATCH] fix: address bug detection findings in token counter and
 streaming iterator

- token_counter: guard against non-dict 'function' field in tool dicts
  and skip tools missing a name to avoid emitting 'type None = ...' which
  would produce inaccurate token counts.
- streaming_iterator: change sync __next__ generic-error path to raise
  StopIteration (was StopAsyncIteration), so sync iteration cleanly stops.
- streaming_iterator: centralize context_management attachment so the
  held-stop_reason direct-flush path defensively re-attaches applied_edits
  to match the merge path's guarantee.

Co-authored-by: Yassin Kortam <yassin@berri.ai>
---
 litellm/litellm_core_utils/token_counter.py   |  8 +++--
 .../adapters/streaming_iterator.py            | 31 +++++++++++++++++--
 2 files changed, 35 insertions(+), 4 deletions(-)

diff --git a/litellm/litellm_core_utils/token_counter.py b/litellm/litellm_core_utils/token_counter.py
index 819bdaf7645..b0dcd9107e3 100644
--- a/litellm/litellm_core_utils/token_counter.py
+++ b/litellm/litellm_core_utils/token_counter.py
@@ -767,7 +767,7 @@ def _format_function_definitions(tools):
         if not isinstance(tool, dict):
             continue
         function = tool.get("function")
-        if function is None:
+        if not isinstance(function, dict):
             # Anthropic tool shape → OpenAI function dict for token counting.
             params = tool.get("input_schema") or tool.get("parameters") or {}
             if not isinstance(params, dict):
@@ -777,9 +777,13 @@ def _format_function_definitions(tools):
                 "description": tool.get("description"),
                 "parameters": params,
             }
+        function_name = function.get("name")
+        if not function_name:
+            # Skip malformed tools missing a name to avoid emitting
+            # ``type None = ...`` which would produce inaccurate token counts.
+            continue
         if function_description := function.get("description"):
             lines.append(f"// {function_description}")
-        function_name = function.get("name")
         parameters = function.get("parameters") or {}
         if not isinstance(parameters, dict):
             parameters = {}
diff --git a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py
index 3a0775e4b56..ef4b3314e31 100644
--- a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py
+++ b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py
@@ -130,10 +130,37 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper):
             )
         return self._augment_message_delta_usage(merged_chunk)
 
+    def _ensure_context_management_attached(
+        self, message_delta_chunk: Dict[str, Any]
+    ) -> Dict[str, Any]:
+        """Attach ``context_management`` to a ``message_delta`` chunk if
+        ``self.applied_edits`` is non-empty and the chunk does not already
+        carry it. Returns the (possibly new) chunk dict.
+
+        Centralizing this guard ensures every ``message_delta`` emission
+        path (merge-with-usage and direct-flush-of-held) consistently
+        surfaces ``applied_edits`` to the client.
+        """
+        if not self.applied_edits or "context_management" in message_delta_chunk:
+            return message_delta_chunk
+        augmented = message_delta_chunk.copy()
+        augmented["context_management"] = ContextManagementResponse(
+            applied_edits=list(self.applied_edits)
+        )
+        return augmented
+
     def _augment_message_delta_usage(
         self, message_delta_chunk: Dict[str, Any]
     ) -> Dict[str, Any]:
-        """Attach polyfill compaction iteration usage to the final message_delta."""
+        """Attach polyfill compaction iteration usage to the final message_delta.
+
+        Also defensively re-attaches ``context_management`` so the direct
+        held-chunk flush path stays in sync with the merge path's guarantee
+        when ``self.applied_edits`` is non-empty.
+        """
+        message_delta_chunk = self._ensure_context_management_attached(
+            message_delta_chunk
+        )
         if self.iterations_usage is None:
             return message_delta_chunk
         usage = message_delta_chunk.get("usage")
@@ -433,7 +460,7 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper):
             verbose_logger.error(
                 "Anthropic Adapter - {}\n{}".format(e, traceback.format_exc())
             )
-            raise StopAsyncIteration
+            raise StopIteration
 
     async def __anext__(self):  # noqa: PLR0915
         from .transformation import LiteLLMAnthropicMessagesAdapter