Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.
Add call_id to LLM events for correlating requests (#4281)
Vini Brasil576b74b2ef3e
diff --git a/lib/crewai/src/crewai/events/types/llm_events.py b/lib/crewai/src/crewai/events/types/llm_events.pyindex 161b8a2a0..87087f100 100644--- a/lib/crewai/src/crewai/events/types/llm_events.py+++ b/lib/crewai/src/crewai/events/types/llm_events.py@@ -10,6 +10,7 @@ class LLMEventBase(BaseEvent):from_task: Any | None = Nonefrom_agent: Any | None = Nonemodel: str | None = None+ call_id: strdef __init__(self, **data: Any) -> None:if data.get("from_task"):diff --git a/lib/crewai/src/crewai/llm.py b/lib/crewai/src/crewai/llm.pyindex c607f1615..902a3d310 100644--- a/lib/crewai/src/crewai/llm.py+++ b/lib/crewai/src/crewai/llm.py@@ -37,7 +37,7 @@ from crewai.events.types.tool_usage_events import (ToolUsageFinishedEvent,ToolUsageStartedEvent,)-from crewai.llms.base_llm import BaseLLM+from crewai.llms.base_llm import BaseLLM, get_current_call_id, llm_call_contextfrom crewai.llms.constants import (ANTHROPIC_MODELS,AZURE_MODELS,@@ -770,7 +770,7 @@ class LLM(BaseLLM):chunk_content = Noneresponse_id = None- if hasattr(chunk,'id'):+ if hasattr(chunk, "id"):response_id = chunk.id# Safely extract content from various chunk formats@@ -827,7 +827,7 @@ class LLM(BaseLLM):available_functions=available_functions,from_task=from_task,from_agent=from_agent,- response_id=response_id+ response_id=response_id,)if result is not None:@@ -849,7 +849,8 @@ class LLM(BaseLLM):from_task=from_task,from_agent=from_agent,call_type=LLMCallType.LLM_CALL,- response_id=response_id+ response_id=response_id,+ call_id=get_current_call_id(),),)# --- 4) Fallback to non-streaming if no content received@@ -1015,7 +1016,10 @@ class LLM(BaseLLM):crewai_event_bus.emit(self,event=LLMCallFailedEvent(- error=str(e), from_task=from_task, from_agent=from_agent+ error=str(e),+ from_task=from_task,+ from_agent=from_agent,+ call_id=get_current_call_id(),),)raise Exception(f"Failed to get streaming response: {e!s}") from e@@ -1048,7 +1052,8 @@ class LLM(BaseLLM):from_task=from_task,from_agent=from_agent,call_type=LLMCallType.TOOL_CALL,- response_id=response_id+ response_id=response_id,+ call_id=get_current_call_id(),),)@@ -1476,7 +1481,8 @@ class LLM(BaseLLM):chunk=chunk_content,from_task=from_task,from_agent=from_agent,- response_id=response_id+ response_id=response_id,+ call_id=get_current_call_id(),),)@@ -1619,7 +1625,12 @@ class LLM(BaseLLM):logging.error(f"Error executing function '{function_name}': {e}")crewai_event_bus.emit(self,- event=LLMCallFailedEvent(error=f"Tool execution error: {e!s}"),+ event=LLMCallFailedEvent(+ error=f"Tool execution error: {e!s}",+ from_task=from_task,+ from_agent=from_agent,+ call_id=get_current_call_id(),+ ),)crewai_event_bus.emit(self,@@ -1669,108 +1680,117 @@ class LLM(BaseLLM):ValueError: If response format is not supportedLLMContextLengthExceededError: If input exceeds model's context limit"""- crewai_event_bus.emit(- self,- event=LLMCallStartedEvent(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- model=self.model,- ),- )-- # --- 2) Validate parameters before proceeding with the call- self._validate_call_params()-- # --- 3) Convert string messages to proper format if needed- if isinstance(messages, str):- messages = [{"role": "user", "content": messages}]- # --- 4) Handle O1 model special case (system messages not supported)- if "o1" in self.model.lower():- for message in messages:- if message.get("role") == "system":- msg_role: Literal["assistant"] = "assistant"- message["role"] = msg_role-- if not self._invoke_before_llm_call_hooks(messages, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")-- # --- 5) Set up callbacks if provided- with suppress_warnings():- if callbacks and len(callbacks) > 0:- self.set_callbacks(callbacks)- try:- # --- 6) Prepare parameters for the completion call- params = self._prepare_completion_params(messages, tools)- # --- 7) Make the completion call and handle response- if self.stream:- result = self._handle_streaming_response(- params=params,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )- else:- result = self._handle_non_streaming_response(- params=params,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )-- if isinstance(result, str):- result = self._invoke_after_llm_call_hooks(- messages, result, from_agent- )+ with llm_call_context() as call_id:+ crewai_event_bus.emit(+ self,+ event=LLMCallStartedEvent(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ model=self.model,+ call_id=call_id,+ ),+ )- return result- except LLMContextLengthExceededError:- # Re-raise LLMContextLengthExceededError as it should be handled- # by the CrewAgentExecutor._invoke_loop method, which can then decide- # whether to summarize the content or abort based on the respect_context_window flag- raise- except Exception as e:- unsupported_stop = "Unsupported parameter" in str(- e- ) and "'stop'" in str(e)-- if unsupported_stop:- if (- "additional_drop_params" in self.additional_params- and isinstance(- self.additional_params["additional_drop_params"], list+ # --- 2) Validate parameters before proceeding with the call+ self._validate_call_params()++ # --- 3) Convert string messages to proper format if needed+ if isinstance(messages, str):+ messages = [{"role": "user", "content": messages}]+ # --- 4) Handle O1 model special case (system messages not supported)+ if "o1" in self.model.lower():+ for message in messages:+ if message.get("role") == "system":+ msg_role: Literal["assistant"] = "assistant"+ message["role"] = msg_role++ if not self._invoke_before_llm_call_hooks(messages, from_agent):+ raise ValueError("LLM call blocked by before_llm_call hook")++ # --- 5) Set up callbacks if provided+ with suppress_warnings():+ if callbacks and len(callbacks) > 0:+ self.set_callbacks(callbacks)+ try:+ # --- 6) Prepare parameters for the completion call+ params = self._prepare_completion_params(messages, tools)+ # --- 7) Make the completion call and handle response+ if self.stream:+ result = self._handle_streaming_response(+ params=params,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,)- ):- self.additional_params["additional_drop_params"].append("stop")else:- self.additional_params = {"additional_drop_params": ["stop"]}+ result = self._handle_non_streaming_response(+ params=params,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,+ )- logging.info("Retrying LLM call without the unsupported 'stop'")+ if isinstance(result, str):+ result = self._invoke_after_llm_call_hooks(+ messages, result, from_agent+ )- return self.call(- messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )+ return result+ except LLMContextLengthExceededError:+ # Re-raise LLMContextLengthExceededError as it should be handled+ # by the CrewAgentExecutor._invoke_loop method, which can then decide+ # whether to summarize the content or abort based on the respect_context_window flag+ raise+ except Exception as e:+ unsupported_stop = "Unsupported parameter" in str(+ e+ ) and "'stop'" in str(e)++ if unsupported_stop:+ if (+ "additional_drop_params" in self.additional_params+ and isinstance(+ self.additional_params["additional_drop_params"], list+ )+ ):+ self.additional_params["additional_drop_params"].append(+ "stop"+ )+ else:+ self.additional_params = {+ "additional_drop_params": ["stop"]+ }++ logging.info("Retrying LLM call without the unsupported 'stop'")++ return self.call(+ messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,+ )- crewai_event_bus.emit(- self,- event=LLMCallFailedEvent(- error=str(e), from_task=from_task, from_agent=from_agent- ),- )- raise+ crewai_event_bus.emit(+ self,+ event=LLMCallFailedEvent(+ error=str(e),+ from_task=from_task,+ from_agent=from_agent,+ call_id=get_current_call_id(),+ ),+ )+ raiseasync def acall(self,@@ -1808,96 +1828,105 @@ class LLM(BaseLLM):ValueError: If response format is not supportedLLMContextLengthExceededError: If input exceeds model's context limit"""- crewai_event_bus.emit(- self,- event=LLMCallStartedEvent(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- model=self.model,- ),- )-- self._validate_call_params()+ with llm_call_context() as call_id:+ crewai_event_bus.emit(+ self,+ event=LLMCallStartedEvent(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ model=self.model,+ call_id=call_id,+ ),+ )- if isinstance(messages, str):- messages = [{"role": "user", "content": messages}]+ self._validate_call_params()- # Process file attachments asynchronously before preparing params- messages = await self._aprocess_message_files(messages)+ if isinstance(messages, str):+ messages = [{"role": "user", "content": messages}]- if "o1" in self.model.lower():- for message in messages:- if message.get("role") == "system":- msg_role: Literal["assistant"] = "assistant"- message["role"] = msg_role+ # Process file attachments asynchronously before preparing params+ messages = await self._aprocess_message_files(messages)- with suppress_warnings():- if callbacks and len(callbacks) > 0:- self.set_callbacks(callbacks)- try:- params = self._prepare_completion_params(- messages, tools, skip_file_processing=True- )+ if "o1" in self.model.lower():+ for message in messages:+ if message.get("role") == "system":+ msg_role: Literal["assistant"] = "assistant"+ message["role"] = msg_role- if self.stream:- return await self._ahandle_streaming_response(- params=params,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,+ with suppress_warnings():+ if callbacks and len(callbacks) > 0:+ self.set_callbacks(callbacks)+ try:+ params = self._prepare_completion_params(+ messages, tools, skip_file_processing=True)- return await self._ahandle_non_streaming_response(- params=params,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )- except LLMContextLengthExceededError:- raise- except Exception as e:- unsupported_stop = "Unsupported parameter" in str(- e- ) and "'stop'" in str(e)-- if unsupported_stop:- if (- "additional_drop_params" in self.additional_params- and isinstance(- self.additional_params["additional_drop_params"], list+ if self.stream:+ return await self._ahandle_streaming_response(+ params=params,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,)- ):- self.additional_params["additional_drop_params"].append("stop")- else:- self.additional_params = {"additional_drop_params": ["stop"]}-- logging.info("Retrying LLM call without the unsupported 'stop'")- return await self.acall(- messages,- tools=tools,+ return await self._ahandle_non_streaming_response(+ params=params,callbacks=callbacks,available_functions=available_functions,from_task=from_task,from_agent=from_agent,response_model=response_model,)+ except LLMContextLengthExceededError:+ raise+ except Exception as e:+ unsupported_stop = "Unsupported parameter" in str(+ e+ ) and "'stop'" in str(e)++ if unsupported_stop:+ if (+ "additional_drop_params" in self.additional_params+ and isinstance(+ self.additional_params["additional_drop_params"], list+ )+ ):+ self.additional_params["additional_drop_params"].append(+ "stop"+ )+ else:+ self.additional_params = {+ "additional_drop_params": ["stop"]+ }++ logging.info("Retrying LLM call without the unsupported 'stop'")++ return await self.acall(+ messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,+ )- crewai_event_bus.emit(- self,- event=LLMCallFailedEvent(- error=str(e), from_task=from_task, from_agent=from_agent- ),- )- raise+ crewai_event_bus.emit(+ self,+ event=LLMCallFailedEvent(+ error=str(e),+ from_task=from_task,+ from_agent=from_agent,+ call_id=get_current_call_id(),+ ),+ )+ raisedef _handle_emit_call_events(self,@@ -1925,6 +1954,7 @@ class LLM(BaseLLM):from_task=from_task,from_agent=from_agent,model=self.model,+ call_id=get_current_call_id(),),)diff --git a/lib/crewai/src/crewai/llms/base_llm.py b/lib/crewai/src/crewai/llms/base_llm.pyindex 56e6dcb34..dcb261fd7 100644--- a/lib/crewai/src/crewai/llms/base_llm.py+++ b/lib/crewai/src/crewai/llms/base_llm.py@@ -7,11 +7,15 @@ in CrewAI, including common functionality for native SDK implementations.from __future__ import annotationsfrom abc import ABC, abstractmethod+from collections.abc import Generator+from contextlib import contextmanager+import contextvarsfrom datetime import datetimeimport jsonimport loggingimport refrom typing import TYPE_CHECKING, Any, Final+import uuidfrom pydantic import BaseModel@@ -50,6 +54,32 @@ DEFAULT_CONTEXT_WINDOW_SIZE: Final[int] = 4096DEFAULT_SUPPORTS_STOP_WORDS: Final[bool] = True_JSON_EXTRACTION_PATTERN: Final[re.Pattern[str]] = re.compile(r"\{.*}", re.DOTALL)+_current_call_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(+ "_current_call_id", default=None+)+++@contextmanager+def llm_call_context() -> Generator[str, None, None]:+ """Context manager that establishes an LLM call scope with a unique call_id."""+ call_id = str(uuid.uuid4())+ token = _current_call_id.set(call_id)+ try:+ yield call_id+ finally:+ _current_call_id.reset(token)+++def get_current_call_id() -> str:+ """Get current call_id from context"""+ call_id = _current_call_id.get()+ if call_id is None:+ logging.warning(+ "LLM event emitted outside call context - generating fallback call_id"+ )+ return str(uuid.uuid4())+ return call_id+class BaseLLM(ABC):"""Abstract base class for LLM implementations.@@ -351,6 +381,7 @@ class BaseLLM(ABC):from_task=from_task,from_agent=from_agent,model=self.model,+ call_id=get_current_call_id(),),)@@ -374,6 +405,7 @@ class BaseLLM(ABC):from_task=from_task,from_agent=from_agent,model=self.model,+ call_id=get_current_call_id(),),)@@ -394,6 +426,7 @@ class BaseLLM(ABC):from_task=from_task,from_agent=from_agent,model=self.model,+ call_id=get_current_call_id(),),)@@ -428,6 +461,7 @@ class BaseLLM(ABC):from_agent=from_agent,call_type=call_type,response_id=response_id,+ call_id=get_current_call_id(),),)diff --git a/lib/crewai/src/crewai/llms/providers/anthropic/completion.py b/lib/crewai/src/crewai/llms/providers/anthropic/completion.pyindex c2ea752c6..815dfe763 100644--- a/lib/crewai/src/crewai/llms/providers/anthropic/completion.py+++ b/lib/crewai/src/crewai/llms/providers/anthropic/completion.py@@ -8,7 +8,7 @@ from typing import TYPE_CHECKING, Any, Final, Literal, TypeGuard, castfrom pydantic import BaseModelfrom crewai.events.types.llm_events import LLMCallType-from crewai.llms.base_llm import BaseLLM+from crewai.llms.base_llm import BaseLLM, llm_call_contextfrom crewai.llms.hooks.transport import AsyncHTTPTransport, HTTPTransportfrom crewai.utilities.agent_utils import is_context_length_exceededfrom crewai.utilities.exceptions.context_window_exceeding_exception import (@@ -266,35 +266,46 @@ class AnthropicCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- # Emit call started event- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ # Emit call started event+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- # Format messages for Anthropic- formatted_messages, system_message = self._format_messages_for_anthropic(- messages- )+ # Format messages for Anthropic+ formatted_messages, system_message = (+ self._format_messages_for_anthropic(messages)+ )- if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")+ if not self._invoke_before_llm_call_hooks(+ formatted_messages, from_agent+ ):+ raise ValueError("LLM call blocked by before_llm_call hook")- # Prepare completion parameters- completion_params = self._prepare_completion_params(- formatted_messages, system_message, tools- )+ # Prepare completion parameters+ completion_params = self._prepare_completion_params(+ formatted_messages, system_message, tools+ )- effective_response_model = response_model or self.response_format+ effective_response_model = response_model or self.response_format++ # Handle streaming vs non-streaming+ if self.stream:+ return self._handle_streaming_completion(+ completion_params,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- # Handle streaming vs non-streaming- if self.stream:- return self._handle_streaming_completion(+ return self._handle_completion(completion_params,available_functions,from_task,@@ -302,21 +313,13 @@ class AnthropicCompletion(BaseLLM):effective_response_model,)- return self._handle_completion(- completion_params,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except Exception as e:- error_msg = f"Anthropic API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except Exception as e:+ error_msg = f"Anthropic API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raiseasync def acall(self,@@ -342,28 +345,37 @@ class AnthropicCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- formatted_messages, system_message = self._format_messages_for_anthropic(- messages- )+ formatted_messages, system_message = (+ self._format_messages_for_anthropic(messages)+ )- completion_params = self._prepare_completion_params(- formatted_messages, system_message, tools- )+ completion_params = self._prepare_completion_params(+ formatted_messages, system_message, tools+ )- effective_response_model = response_model or self.response_format+ effective_response_model = response_model or self.response_format++ if self.stream:+ return await self._ahandle_streaming_completion(+ completion_params,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- if self.stream:- return await self._ahandle_streaming_completion(+ return await self._ahandle_completion(completion_params,available_functions,from_task,@@ -371,21 +383,13 @@ class AnthropicCompletion(BaseLLM):effective_response_model,)- return await self._ahandle_completion(- completion_params,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except Exception as e:- error_msg = f"Anthropic API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except Exception as e:+ error_msg = f"Anthropic API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raisedef _prepare_completion_params(self,diff --git a/lib/crewai/src/crewai/llms/providers/azure/completion.py b/lib/crewai/src/crewai/llms/providers/azure/completion.pyindex 8f398a594..e7fd80844 100644--- a/lib/crewai/src/crewai/llms/providers/azure/completion.py+++ b/lib/crewai/src/crewai/llms/providers/azure/completion.py@@ -43,7 +43,7 @@ try:)from crewai.events.types.llm_events import LLMCallType- from crewai.llms.base_llm import BaseLLM+ from crewai.llms.base_llm import BaseLLM, llm_call_contextexcept ImportError:raise ImportError(@@ -293,32 +293,44 @@ class AzureCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- # Emit call started event- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )- effective_response_model = response_model or self.response_format+ with llm_call_context():+ try:+ # Emit call started event+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- # Format messages for Azure- formatted_messages = self._format_messages_for_azure(messages)+ effective_response_model = response_model or self.response_format- if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")+ # Format messages for Azure+ formatted_messages = self._format_messages_for_azure(messages)- # Prepare completion parameters- completion_params = self._prepare_completion_params(- formatted_messages, tools, effective_response_model- )+ if not self._invoke_before_llm_call_hooks(+ formatted_messages, from_agent+ ):+ raise ValueError("LLM call blocked by before_llm_call hook")++ # Prepare completion parameters+ completion_params = self._prepare_completion_params(+ formatted_messages, tools, effective_response_model+ )++ # Handle streaming vs non-streaming+ if self.stream:+ return self._handle_streaming_completion(+ completion_params,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- # Handle streaming vs non-streaming- if self.stream:- return self._handle_streaming_completion(+ return self._handle_completion(completion_params,available_functions,from_task,@@ -326,16 +338,8 @@ class AzureCompletion(BaseLLM):effective_response_model,)- return self._handle_completion(- completion_params,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except Exception as e:- return self._handle_api_error(e, from_task, from_agent) # type: ignore[func-returns-value]+ except Exception as e:+ return self._handle_api_error(e, from_task, from_agent) # type: ignore[func-returns-value]async def acall( # type: ignore[return]self,@@ -361,25 +365,35 @@ class AzureCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )- effective_response_model = response_model or self.response_format+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- formatted_messages = self._format_messages_for_azure(messages)+ effective_response_model = response_model or self.response_format- completion_params = self._prepare_completion_params(- formatted_messages, tools, effective_response_model- )+ formatted_messages = self._format_messages_for_azure(messages)++ completion_params = self._prepare_completion_params(+ formatted_messages, tools, effective_response_model+ )++ if self.stream:+ return await self._ahandle_streaming_completion(+ completion_params,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- if self.stream:- return await self._ahandle_streaming_completion(+ return await self._ahandle_completion(completion_params,available_functions,from_task,@@ -387,16 +401,8 @@ class AzureCompletion(BaseLLM):effective_response_model,)- return await self._ahandle_completion(- completion_params,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except Exception as e:- self._handle_api_error(e, from_task, from_agent)+ except Exception as e:+ self._handle_api_error(e, from_task, from_agent)def _prepare_completion_params(self,diff --git a/lib/crewai/src/crewai/llms/providers/bedrock/completion.py b/lib/crewai/src/crewai/llms/providers/bedrock/completion.pyindex 004b076f8..47946d949 100644--- a/lib/crewai/src/crewai/llms/providers/bedrock/completion.py+++ b/lib/crewai/src/crewai/llms/providers/bedrock/completion.py@@ -11,7 +11,7 @@ from pydantic import BaseModelfrom typing_extensions import Requiredfrom crewai.events.types.llm_events import LLMCallType-from crewai.llms.base_llm import BaseLLM+from crewai.llms.base_llm import BaseLLM, llm_call_contextfrom crewai.utilities.agent_utils import is_context_length_exceededfrom crewai.utilities.exceptions.context_window_exceeding_exception import (LLMContextLengthExceededError,@@ -378,77 +378,90 @@ class BedrockCompletion(BaseLLM):"""Call AWS Bedrock Converse API."""effective_response_model = response_model or self.response_format- try:- # Emit call started event- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ # Emit call started event+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- # Format messages for Converse API- formatted_messages, system_message = self._format_messages_for_converse(- messages- )+ # Format messages for Converse API+ formatted_messages, system_message = self._format_messages_for_converse(+ messages+ )- if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")+ if not self._invoke_before_llm_call_hooks(+ formatted_messages, from_agent+ ):+ raise ValueError("LLM call blocked by before_llm_call hook")- # Prepare request body- body: BedrockConverseRequestBody = {- "inferenceConfig": self._get_inference_config(),- }+ # Prepare request body+ body: BedrockConverseRequestBody = {+ "inferenceConfig": self._get_inference_config(),+ }- # Add system message if present- if system_message:- body["system"] = cast(- "list[SystemContentBlockTypeDef]",- cast(object, [{"text": system_message}]),- )+ # Add system message if present+ if system_message:+ body["system"] = cast(+ "list[SystemContentBlockTypeDef]",+ cast(object, [{"text": system_message}]),+ )- # Add tool config if present or if messages contain tool content- # Bedrock requires toolConfig when messages have toolUse/toolResult- if tools:- tool_config: ToolConfigurationTypeDef = {- "tools": cast(- "Sequence[ToolTypeDef]",- cast(object, self._format_tools_for_converse(tools)),+ # Add tool config if present or if messages contain tool content+ # Bedrock requires toolConfig when messages have toolUse/toolResult+ if tools:+ tool_config: ToolConfigurationTypeDef = {+ "tools": cast(+ "Sequence[ToolTypeDef]",+ cast(object, self._format_tools_for_converse(tools)),+ )+ }+ body["toolConfig"] = tool_config+ elif self._messages_contain_tool_content(formatted_messages):+ # Create minimal toolConfig from tool history in messages+ tools_from_history = self._extract_tools_from_message_history(+ formatted_messages)- }- body["toolConfig"] = tool_config- elif self._messages_contain_tool_content(formatted_messages):- # Create minimal toolConfig from tool history in messages- tools_from_history = self._extract_tools_from_message_history(- formatted_messages- )- if tools_from_history:- body["toolConfig"] = cast(- "ToolConfigurationTypeDef",- cast(object, {"tools": tools_from_history}),+ if tools_from_history:+ body["toolConfig"] = cast(+ "ToolConfigurationTypeDef",+ cast(object, {"tools": tools_from_history}),+ )++ # Add optional advanced features if configured+ if self.guardrail_config:+ guardrail_config: GuardrailConfigurationTypeDef = cast(+ "GuardrailConfigurationTypeDef",+ cast(object, self.guardrail_config),)+ body["guardrailConfig"] = guardrail_config- # Add optional advanced features if configured- if self.guardrail_config:- guardrail_config: GuardrailConfigurationTypeDef = cast(- "GuardrailConfigurationTypeDef", cast(object, self.guardrail_config)- )- body["guardrailConfig"] = guardrail_config+ if self.additional_model_request_fields:+ body["additionalModelRequestFields"] = (+ self.additional_model_request_fields+ )- if self.additional_model_request_fields:- body["additionalModelRequestFields"] = (- self.additional_model_request_fields- )+ if self.additional_model_response_field_paths:+ body["additionalModelResponseFieldPaths"] = (+ self.additional_model_response_field_paths+ )- if self.additional_model_response_field_paths:- body["additionalModelResponseFieldPaths"] = (- self.additional_model_response_field_paths- )+ if self.stream:+ return self._handle_streaming_converse(+ formatted_messages,+ body,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- if self.stream:- return self._handle_streaming_converse(+ return self._handle_converse(formatted_messages,body,available_functions,@@ -457,26 +470,17 @@ class BedrockCompletion(BaseLLM):effective_response_model,)- return self._handle_converse(- formatted_messages,- body,- available_functions,- from_task,- from_agent,- effective_response_model,- )+ except Exception as e:+ if is_context_length_exceeded(e):+ logging.error(f"Context window exceeded: {e}")+ raise LLMContextLengthExceededError(str(e)) from e- except Exception as e:- if is_context_length_exceeded(e):- logging.error(f"Context window exceeded: {e}")- raise LLMContextLengthExceededError(str(e)) from e-- error_msg = f"AWS Bedrock API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ error_msg = f"AWS Bedrock API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raiseasync def acall(self,@@ -514,69 +518,80 @@ class BedrockCompletion(BaseLLM):'Install with: uv add "crewai[bedrock-async]"')- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )- formatted_messages, system_message = self._format_messages_for_converse(- messages- )+ formatted_messages, system_message = self._format_messages_for_converse(+ messages+ )- body: BedrockConverseRequestBody = {- "inferenceConfig": self._get_inference_config(),- }+ body: BedrockConverseRequestBody = {+ "inferenceConfig": self._get_inference_config(),+ }- if system_message:- body["system"] = cast(- "list[SystemContentBlockTypeDef]",- cast(object, [{"text": system_message}]),- )+ if system_message:+ body["system"] = cast(+ "list[SystemContentBlockTypeDef]",+ cast(object, [{"text": system_message}]),+ )- # Add tool config if present or if messages contain tool content- # Bedrock requires toolConfig when messages have toolUse/toolResult- if tools:- tool_config: ToolConfigurationTypeDef = {- "tools": cast(- "Sequence[ToolTypeDef]",- cast(object, self._format_tools_for_converse(tools)),+ # Add tool config if present or if messages contain tool content+ # Bedrock requires toolConfig when messages have toolUse/toolResult+ if tools:+ tool_config: ToolConfigurationTypeDef = {+ "tools": cast(+ "Sequence[ToolTypeDef]",+ cast(object, self._format_tools_for_converse(tools)),+ )+ }+ body["toolConfig"] = tool_config+ elif self._messages_contain_tool_content(formatted_messages):+ # Create minimal toolConfig from tool history in messages+ tools_from_history = self._extract_tools_from_message_history(+ formatted_messages)- }- body["toolConfig"] = tool_config- elif self._messages_contain_tool_content(formatted_messages):- # Create minimal toolConfig from tool history in messages- tools_from_history = self._extract_tools_from_message_history(- formatted_messages- )- if tools_from_history:- body["toolConfig"] = cast(- "ToolConfigurationTypeDef",- cast(object, {"tools": tools_from_history}),+ if tools_from_history:+ body["toolConfig"] = cast(+ "ToolConfigurationTypeDef",+ cast(object, {"tools": tools_from_history}),+ )++ if self.guardrail_config:+ guardrail_config: GuardrailConfigurationTypeDef = cast(+ "GuardrailConfigurationTypeDef",+ cast(object, self.guardrail_config),)+ body["guardrailConfig"] = guardrail_config- if self.guardrail_config:- guardrail_config: GuardrailConfigurationTypeDef = cast(- "GuardrailConfigurationTypeDef", cast(object, self.guardrail_config)- )- body["guardrailConfig"] = guardrail_config+ if self.additional_model_request_fields:+ body["additionalModelRequestFields"] = (+ self.additional_model_request_fields+ )- if self.additional_model_request_fields:- body["additionalModelRequestFields"] = (- self.additional_model_request_fields- )+ if self.additional_model_response_field_paths:+ body["additionalModelResponseFieldPaths"] = (+ self.additional_model_response_field_paths+ )- if self.additional_model_response_field_paths:- body["additionalModelResponseFieldPaths"] = (- self.additional_model_response_field_paths- )+ if self.stream:+ return await self._ahandle_streaming_converse(+ formatted_messages,+ body,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )- if self.stream:- return await self._ahandle_streaming_converse(+ return await self._ahandle_converse(formatted_messages,body,available_functions,@@ -585,26 +600,17 @@ class BedrockCompletion(BaseLLM):effective_response_model,)- return await self._ahandle_converse(- formatted_messages,- body,- available_functions,- from_task,- from_agent,- effective_response_model,- )+ except Exception as e:+ if is_context_length_exceeded(e):+ logging.error(f"Context window exceeded: {e}")+ raise LLMContextLengthExceededError(str(e)) from e- except Exception as e:- if is_context_length_exceeded(e):- logging.error(f"Context window exceeded: {e}")- raise LLMContextLengthExceededError(str(e)) from e-- error_msg = f"AWS Bedrock API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ error_msg = f"AWS Bedrock API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raisedef _handle_converse(self,diff --git a/lib/crewai/src/crewai/llms/providers/gemini/completion.py b/lib/crewai/src/crewai/llms/providers/gemini/completion.pyindex 4a28639f5..0c00de96d 100644--- a/lib/crewai/src/crewai/llms/providers/gemini/completion.py+++ b/lib/crewai/src/crewai/llms/providers/gemini/completion.py@@ -10,7 +10,7 @@ from typing import TYPE_CHECKING, Any, Literal, castfrom pydantic import BaseModelfrom crewai.events.types.llm_events import LLMCallType-from crewai.llms.base_llm import BaseLLM+from crewai.llms.base_llm import BaseLLM, llm_call_contextfrom crewai.utilities.agent_utils import is_context_length_exceededfrom crewai.utilities.exceptions.context_window_exceeding_exception import (LLMContextLengthExceededError,@@ -293,33 +293,45 @@ class GeminiCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )- self.tools = tools- effective_response_model = response_model or self.response_format+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )+ self.tools = tools+ effective_response_model = response_model or self.response_format- formatted_content, system_instruction = self._format_messages_for_gemini(- messages- )+ formatted_content, system_instruction = (+ self._format_messages_for_gemini(messages)+ )- messages_for_hooks = self._convert_contents_to_dict(formatted_content)+ messages_for_hooks = self._convert_contents_to_dict(formatted_content)- if not self._invoke_before_llm_call_hooks(messages_for_hooks, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")+ if not self._invoke_before_llm_call_hooks(+ messages_for_hooks, from_agent+ ):+ raise ValueError("LLM call blocked by before_llm_call hook")- config = self._prepare_generation_config(- system_instruction, tools, effective_response_model- )+ config = self._prepare_generation_config(+ system_instruction, tools, effective_response_model+ )- if self.stream:- return self._handle_streaming_completion(+ if self.stream:+ return self._handle_streaming_completion(+ formatted_content,+ config,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )++ return self._handle_completion(formatted_content,config,available_functions,@@ -328,29 +340,20 @@ class GeminiCompletion(BaseLLM):effective_response_model,)- return self._handle_completion(- formatted_content,- config,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except APIError as e:- error_msg = f"Google Gemini API error: {e.code} - {e.message}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise- except Exception as e:- error_msg = f"Google Gemini API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except APIError as e:+ error_msg = f"Google Gemini API error: {e.code} - {e.message}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raise+ except Exception as e:+ error_msg = f"Google Gemini API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raiseasync def acall(self,@@ -376,28 +379,38 @@ class GeminiCompletion(BaseLLM):Returns:Chat completion response or tool call result"""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )- self.tools = tools- effective_response_model = response_model or self.response_format+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )+ self.tools = tools+ effective_response_model = response_model or self.response_format- formatted_content, system_instruction = self._format_messages_for_gemini(- messages- )+ formatted_content, system_instruction = (+ self._format_messages_for_gemini(messages)+ )- config = self._prepare_generation_config(- system_instruction, tools, effective_response_model- )+ config = self._prepare_generation_config(+ system_instruction, tools, effective_response_model+ )- if self.stream:- return await self._ahandle_streaming_completion(+ if self.stream:+ return await self._ahandle_streaming_completion(+ formatted_content,+ config,+ available_functions,+ from_task,+ from_agent,+ effective_response_model,+ )++ return await self._ahandle_completion(formatted_content,config,available_functions,@@ -406,29 +419,20 @@ class GeminiCompletion(BaseLLM):effective_response_model,)- return await self._ahandle_completion(- formatted_content,- config,- available_functions,- from_task,- from_agent,- effective_response_model,- )-- except APIError as e:- error_msg = f"Google Gemini API error: {e.code} - {e.message}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise- except Exception as e:- error_msg = f"Google Gemini API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except APIError as e:+ error_msg = f"Google Gemini API error: {e.code} - {e.message}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raise+ except Exception as e:+ error_msg = f"Google Gemini API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raisedef _prepare_generation_config(self,diff --git a/lib/crewai/src/crewai/llms/providers/openai/completion.py b/lib/crewai/src/crewai/llms/providers/openai/completion.pyindex c3c5f6292..37b686e3d 100644--- a/lib/crewai/src/crewai/llms/providers/openai/completion.py+++ b/lib/crewai/src/crewai/llms/providers/openai/completion.py@@ -17,7 +17,7 @@ from openai.types.responses import Responsefrom pydantic import BaseModelfrom crewai.events.types.llm_events import LLMCallType-from crewai.llms.base_llm import BaseLLM+from crewai.llms.base_llm import BaseLLM, llm_call_contextfrom crewai.llms.hooks.transport import AsyncHTTPTransport, HTTPTransportfrom crewai.utilities.agent_utils import is_context_length_exceededfrom crewai.utilities.exceptions.context_window_exceeding_exception import (@@ -382,23 +382,35 @@ class OpenAICompletion(BaseLLM):Returns:Completion response or tool call result."""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )++ formatted_messages = self._format_messages(messages)- formatted_messages = self._format_messages(messages)+ if not self._invoke_before_llm_call_hooks(+ formatted_messages, from_agent+ ):+ raise ValueError("LLM call blocked by before_llm_call hook")- if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):- raise ValueError("LLM call blocked by before_llm_call hook")+ if self.api == "responses":+ return self._call_responses(+ messages=formatted_messages,+ tools=tools,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,+ )- if self.api == "responses":- return self._call_responses(+ return self._call_completions(messages=formatted_messages,tools=tools,available_functions=available_functions,@@ -407,22 +419,13 @@ class OpenAICompletion(BaseLLM):response_model=response_model,)- return self._call_completions(- messages=formatted_messages,- tools=tools,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )-- except Exception as e:- error_msg = f"OpenAI API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except Exception as e:+ error_msg = f"OpenAI API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raisedef _call_completions(self,@@ -479,20 +482,30 @@ class OpenAICompletion(BaseLLM):Returns:Completion response or tool call result."""- try:- self._emit_call_started_event(- messages=messages,- tools=tools,- callbacks=callbacks,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- )+ with llm_call_context():+ try:+ self._emit_call_started_event(+ messages=messages,+ tools=tools,+ callbacks=callbacks,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ )++ formatted_messages = self._format_messages(messages)- formatted_messages = self._format_messages(messages)+ if self.api == "responses":+ return await self._acall_responses(+ messages=formatted_messages,+ tools=tools,+ available_functions=available_functions,+ from_task=from_task,+ from_agent=from_agent,+ response_model=response_model,+ )- if self.api == "responses":- return await self._acall_responses(+ return await self._acall_completions(messages=formatted_messages,tools=tools,available_functions=available_functions,@@ -501,22 +514,13 @@ class OpenAICompletion(BaseLLM):response_model=response_model,)- return await self._acall_completions(- messages=formatted_messages,- tools=tools,- available_functions=available_functions,- from_task=from_task,- from_agent=from_agent,- response_model=response_model,- )-- except Exception as e:- error_msg = f"OpenAI API call failed: {e!s}"- logging.error(error_msg)- self._emit_call_failed_event(- error=error_msg, from_task=from_task, from_agent=from_agent- )- raise+ except Exception as e:+ error_msg = f"OpenAI API call failed: {e!s}"+ logging.error(error_msg)+ self._emit_call_failed_event(+ error=error_msg, from_task=from_task, from_agent=from_agent+ )+ raiseasync def _acall_completions(self,diff --git a/lib/crewai/tests/cassettes/utilities/test_llm_call_events_share_call_id.yaml b/lib/crewai/tests/cassettes/utilities/test_llm_call_events_share_call_id.yamlnew file mode 100644index 000000000..2370a9d04--- /dev/null+++ b/lib/crewai/tests/cassettes/utilities/test_llm_call_events_share_call_id.yaml@@ -0,0 +1,108 @@+interactions:+- request:+ body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini"}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - application/json+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ authorization:+ - AUTHORIZATION-XXX+ connection:+ - keep-alive+ content-length:+ - '71'+ content-type:+ - application/json+ host:+ - api.openai.com+ x-stainless-arch:+ - X-STAINLESS-ARCH-XXX+ x-stainless-async:+ - 'false'+ x-stainless-lang:+ - python+ x-stainless-os:+ - X-STAINLESS-OS-XXX+ x-stainless-package-version:+ - 1.83.0+ x-stainless-read-timeout:+ - X-STAINLESS-READ-TIMEOUT-XXX+ x-stainless-retry-count:+ - '0'+ x-stainless-runtime:+ - CPython+ x-stainless-runtime-version:+ - 3.13.0+ method: POST+ uri: https://api.openai.com/v1/chat/completions+ response:+ body:+ string: "{\n \"id\": \"chatcmpl-D2HpUSxS5LeHwDTELElWlC5CDMzmr\",\n \"object\":+ \"chat.completion\",\n \"created\": 1769437564,\n \"model\": \"gpt-4o-mini-2024-07-18\",\n+ \ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":+ \"assistant\",\n \"content\": \"Hi there! How can I assist you today?\",\n+ \ \"refusal\": null,\n \"annotations\": []\n },\n \"logprobs\":+ null,\n \"finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":+ 9,\n \"completion_tokens\": 10,\n \"total_tokens\": 19,\n \"prompt_tokens_details\":+ {\n \"cached_tokens\": 0,\n \"audio_tokens\": 0\n },\n \"completion_tokens_details\":+ {\n \"reasoning_tokens\": 0,\n \"audio_tokens\": 0,\n \"accepted_prediction_tokens\":+ 0,\n \"rejected_prediction_tokens\": 0\n }\n },\n \"service_tier\":+ \"default\",\n \"system_fingerprint\": \"fp_29330a9688\"\n}\n"+ headers:+ CF-RAY:+ - CF-RAY-XXX+ Connection:+ - keep-alive+ Content-Type:+ - application/json+ Date:+ - Mon, 26 Jan 2026 14:26:05 GMT+ Server:+ - cloudflare+ Set-Cookie:+ - SET-COOKIE-XXX+ Strict-Transport-Security:+ - STS-XXX+ Transfer-Encoding:+ - chunked+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ access-control-expose-headers:+ - ACCESS-CONTROL-XXX+ alt-svc:+ - h3=":443"; ma=86400+ cf-cache-status:+ - DYNAMIC+ openai-organization:+ - OPENAI-ORG-XXX+ openai-processing-ms:+ - '460'+ openai-project:+ - OPENAI-PROJECT-XXX+ openai-version:+ - '2020-10-01'+ x-envoy-upstream-service-time:+ - '477'+ x-openai-proxy-wasm:+ - v0.1+ x-ratelimit-limit-requests:+ - X-RATELIMIT-LIMIT-REQUESTS-XXX+ x-ratelimit-limit-tokens:+ - X-RATELIMIT-LIMIT-TOKENS-XXX+ x-ratelimit-remaining-requests:+ - X-RATELIMIT-REMAINING-REQUESTS-XXX+ x-ratelimit-remaining-tokens:+ - X-RATELIMIT-REMAINING-TOKENS-XXX+ x-ratelimit-reset-requests:+ - X-RATELIMIT-RESET-REQUESTS-XXX+ x-ratelimit-reset-tokens:+ - X-RATELIMIT-RESET-TOKENS-XXX+ x-request-id:+ - X-REQUEST-ID-XXX+ status:+ code: 200+ message: OK+version: 1diff --git a/lib/crewai/tests/cassettes/utilities/test_separate_llm_calls_have_different_call_ids.yaml b/lib/crewai/tests/cassettes/utilities/test_separate_llm_calls_have_different_call_ids.yamlnew file mode 100644index 000000000..419c5e006--- /dev/null+++ b/lib/crewai/tests/cassettes/utilities/test_separate_llm_calls_have_different_call_ids.yaml@@ -0,0 +1,215 @@+interactions:+- request:+ body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini"}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - application/json+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ authorization:+ - AUTHORIZATION-XXX+ connection:+ - keep-alive+ content-length:+ - '71'+ content-type:+ - application/json+ host:+ - api.openai.com+ x-stainless-arch:+ - X-STAINLESS-ARCH-XXX+ x-stainless-async:+ - 'false'+ x-stainless-lang:+ - python+ x-stainless-os:+ - X-STAINLESS-OS-XXX+ x-stainless-package-version:+ - 1.83.0+ x-stainless-read-timeout:+ - X-STAINLESS-READ-TIMEOUT-XXX+ x-stainless-retry-count:+ - '0'+ x-stainless-runtime:+ - CPython+ x-stainless-runtime-version:+ - 3.13.0+ method: POST+ uri: https://api.openai.com/v1/chat/completions+ response:+ body:+ string: "{\n \"id\": \"chatcmpl-D2HpStmyOpe9DrthWBlDdMZfVMJ1u\",\n \"object\":+ \"chat.completion\",\n \"created\": 1769437562,\n \"model\": \"gpt-4o-mini-2024-07-18\",\n+ \ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":+ \"assistant\",\n \"content\": \"Hi! How can I assist you today?\",\n+ \ \"refusal\": null,\n \"annotations\": []\n },\n \"logprobs\":+ null,\n \"finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":+ 9,\n \"completion_tokens\": 9,\n \"total_tokens\": 18,\n \"prompt_tokens_details\":+ {\n \"cached_tokens\": 0,\n \"audio_tokens\": 0\n },\n \"completion_tokens_details\":+ {\n \"reasoning_tokens\": 0,\n \"audio_tokens\": 0,\n \"accepted_prediction_tokens\":+ 0,\n \"rejected_prediction_tokens\": 0\n }\n },\n \"service_tier\":+ \"default\",\n \"system_fingerprint\": \"fp_29330a9688\"\n}\n"+ headers:+ CF-RAY:+ - CF-RAY-XXX+ Connection:+ - keep-alive+ Content-Type:+ - application/json+ Date:+ - Mon, 26 Jan 2026 14:26:02 GMT+ Server:+ - cloudflare+ Set-Cookie:+ - SET-COOKIE-XXX+ Strict-Transport-Security:+ - STS-XXX+ Transfer-Encoding:+ - chunked+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ access-control-expose-headers:+ - ACCESS-CONTROL-XXX+ alt-svc:+ - h3=":443"; ma=86400+ cf-cache-status:+ - DYNAMIC+ openai-organization:+ - OPENAI-ORG-XXX+ openai-processing-ms:+ - '415'+ openai-project:+ - OPENAI-PROJECT-XXX+ openai-version:+ - '2020-10-01'+ x-envoy-upstream-service-time:+ - '434'+ x-openai-proxy-wasm:+ - v0.1+ x-ratelimit-limit-requests:+ - X-RATELIMIT-LIMIT-REQUESTS-XXX+ x-ratelimit-limit-tokens:+ - X-RATELIMIT-LIMIT-TOKENS-XXX+ x-ratelimit-remaining-requests:+ - X-RATELIMIT-REMAINING-REQUESTS-XXX+ x-ratelimit-remaining-tokens:+ - X-RATELIMIT-REMAINING-TOKENS-XXX+ x-ratelimit-reset-requests:+ - X-RATELIMIT-RESET-REQUESTS-XXX+ x-ratelimit-reset-tokens:+ - X-RATELIMIT-RESET-TOKENS-XXX+ x-request-id:+ - X-REQUEST-ID-XXX+ status:+ code: 200+ message: OK+- request:+ body: '{"messages":[{"role":"user","content":"Say bye"}],"model":"gpt-4o-mini"}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - application/json+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ authorization:+ - AUTHORIZATION-XXX+ connection:+ - keep-alive+ content-length:+ - '72'+ content-type:+ - application/json+ cookie:+ - COOKIE-XXX+ host:+ - api.openai.com+ x-stainless-arch:+ - X-STAINLESS-ARCH-XXX+ x-stainless-async:+ - 'false'+ x-stainless-lang:+ - python+ x-stainless-os:+ - X-STAINLESS-OS-XXX+ x-stainless-package-version:+ - 1.83.0+ x-stainless-read-timeout:+ - X-STAINLESS-READ-TIMEOUT-XXX+ x-stainless-retry-count:+ - '0'+ x-stainless-runtime:+ - CPython+ x-stainless-runtime-version:+ - 3.13.0+ method: POST+ uri: https://api.openai.com/v1/chat/completions+ response:+ body:+ string: "{\n \"id\": \"chatcmpl-D2HpS1DP0Xd3tmWt5PBincVrdU7yw\",\n \"object\":+ \"chat.completion\",\n \"created\": 1769437562,\n \"model\": \"gpt-4o-mini-2024-07-18\",\n+ \ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":+ \"assistant\",\n \"content\": \"Goodbye! If you have more questions+ in the future, feel free to reach out. Have a great day!\",\n \"refusal\":+ null,\n \"annotations\": []\n },\n \"logprobs\": null,\n+ \ \"finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":+ 9,\n \"completion_tokens\": 23,\n \"total_tokens\": 32,\n \"prompt_tokens_details\":+ {\n \"cached_tokens\": 0,\n \"audio_tokens\": 0\n },\n \"completion_tokens_details\":+ {\n \"reasoning_tokens\": 0,\n \"audio_tokens\": 0,\n \"accepted_prediction_tokens\":+ 0,\n \"rejected_prediction_tokens\": 0\n }\n },\n \"service_tier\":+ \"default\",\n \"system_fingerprint\": \"fp_29330a9688\"\n}\n"+ headers:+ CF-RAY:+ - CF-RAY-XXX+ Connection:+ - keep-alive+ Content-Type:+ - application/json+ Date:+ - Mon, 26 Jan 2026 14:26:03 GMT+ Server:+ - cloudflare+ Strict-Transport-Security:+ - STS-XXX+ Transfer-Encoding:+ - chunked+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ access-control-expose-headers:+ - ACCESS-CONTROL-XXX+ alt-svc:+ - h3=":443"; ma=86400+ cf-cache-status:+ - DYNAMIC+ openai-organization:+ - OPENAI-ORG-XXX+ openai-processing-ms:+ - '964'+ openai-project:+ - OPENAI-PROJECT-XXX+ openai-version:+ - '2020-10-01'+ x-envoy-upstream-service-time:+ - '979'+ x-openai-proxy-wasm:+ - v0.1+ x-ratelimit-limit-requests:+ - X-RATELIMIT-LIMIT-REQUESTS-XXX+ x-ratelimit-limit-tokens:+ - X-RATELIMIT-LIMIT-TOKENS-XXX+ x-ratelimit-remaining-requests:+ - X-RATELIMIT-REMAINING-REQUESTS-XXX+ x-ratelimit-remaining-tokens:+ - X-RATELIMIT-REMAINING-TOKENS-XXX+ x-ratelimit-reset-requests:+ - X-RATELIMIT-RESET-REQUESTS-XXX+ x-ratelimit-reset-tokens:+ - X-RATELIMIT-RESET-TOKENS-XXX+ x-request-id:+ - X-REQUEST-ID-XXX+ status:+ code: 200+ message: OK+version: 1diff --git a/lib/crewai/tests/cassettes/utilities/test_streaming_chunks_share_call_id_with_call.yaml b/lib/crewai/tests/cassettes/utilities/test_streaming_chunks_share_call_id_with_call.yamlnew file mode 100644index 000000000..7b04d21a3--- /dev/null+++ b/lib/crewai/tests/cassettes/utilities/test_streaming_chunks_share_call_id_with_call.yaml@@ -0,0 +1,143 @@+interactions:+- request:+ body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini","stream":true,"stream_options":{"include_usage":true}}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - application/json+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ authorization:+ - AUTHORIZATION-XXX+ connection:+ - keep-alive+ content-length:+ - '125'+ content-type:+ - application/json+ host:+ - api.openai.com+ x-stainless-arch:+ - X-STAINLESS-ARCH-XXX+ x-stainless-async:+ - 'false'+ x-stainless-lang:+ - python+ x-stainless-os:+ - X-STAINLESS-OS-XXX+ x-stainless-package-version:+ - 1.83.0+ x-stainless-read-timeout:+ - X-STAINLESS-READ-TIMEOUT-XXX+ x-stainless-retry-count:+ - '0'+ x-stainless-runtime:+ - CPython+ x-stainless-runtime-version:+ - 3.13.0+ method: POST+ uri: https://api.openai.com/v1/chat/completions+ response:+ body:+ string: 'data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"role":"assistant","content":"","refusal":null},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"rVIyGQF2E"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"Hi"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"ZGVqV7ZDm"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"!"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"vnfm7IxlIB"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ How"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"o8F35ZZ"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ can"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"kiBzGe3"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ I"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"cbGT2RWgx"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ assist"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"DtxR"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ you"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"6y6Co8J"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"+ today"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"SZOmm"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"?"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"s9Bc0HqlPg"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{},"logprobs":null,"finish_reason":"stop"}],"usage":null,"obfuscation":"u9aar"}+++ data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[],"usage":{"prompt_tokens":9,"completion_tokens":9,"total_tokens":18,"prompt_tokens_details":{"cached_tokens":0,"audio_tokens":0},"completion_tokens_details":{"reasoning_tokens":0,"audio_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"obfuscation":"5hudm8ySqh39"}+++ data: [DONE]+++ '+ headers:+ CF-RAY:+ - CF-RAY-XXX+ Connection:+ - keep-alive+ Content-Type:+ - text/event-stream; charset=utf-8+ Date:+ - Mon, 26 Jan 2026 14:26:04 GMT+ Server:+ - cloudflare+ Set-Cookie:+ - SET-COOKIE-XXX+ Strict-Transport-Security:+ - STS-XXX+ Transfer-Encoding:+ - chunked+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ access-control-expose-headers:+ - ACCESS-CONTROL-XXX+ alt-svc:+ - h3=":443"; ma=86400+ cf-cache-status:+ - DYNAMIC+ openai-organization:+ - OPENAI-ORG-XXX+ openai-processing-ms:+ - '260'+ openai-project:+ - OPENAI-PROJECT-XXX+ openai-version:+ - '2020-10-01'+ x-envoy-upstream-service-time:+ - '275'+ x-openai-proxy-wasm:+ - v0.1+ x-ratelimit-limit-requests:+ - X-RATELIMIT-LIMIT-REQUESTS-XXX+ x-ratelimit-limit-tokens:+ - X-RATELIMIT-LIMIT-TOKENS-XXX+ x-ratelimit-remaining-requests:+ - X-RATELIMIT-REMAINING-REQUESTS-XXX+ x-ratelimit-remaining-tokens:+ - X-RATELIMIT-REMAINING-TOKENS-XXX+ x-ratelimit-reset-requests:+ - X-RATELIMIT-RESET-REQUESTS-XXX+ x-ratelimit-reset-tokens:+ - X-RATELIMIT-RESET-TOKENS-XXX+ x-request-id:+ - X-REQUEST-ID-XXX+ status:+ code: 200+ message: OK+version: 1diff --git a/lib/crewai/tests/test_streaming.py b/lib/crewai/tests/test_streaming.pyindex 5860755ff..8eb63694e 100644--- a/lib/crewai/tests/test_streaming.py+++ b/lib/crewai/tests/test_streaming.py@@ -217,6 +217,7 @@ class TestCrewKickoffStreaming:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="Hello ",+ call_id="test-call-id",),)crewai_event_bus.emit(@@ -224,6 +225,7 @@ class TestCrewKickoffStreaming:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="World!",+ call_id="test-call-id",),)return mock_output@@ -284,6 +286,7 @@ class TestCrewKickoffStreaming:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="",+ call_id="test-call-id",tool_call=ToolCall(id="call-123",function=FunctionCall(@@ -364,6 +367,7 @@ class TestCrewKickoffStreamingAsync:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="Async ",+ call_id="test-call-id",),)crewai_event_bus.emit(@@ -371,6 +375,7 @@ class TestCrewKickoffStreamingAsync:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="Stream!",+ call_id="test-call-id",),)return mock_output@@ -451,6 +456,7 @@ class TestFlowKickoffStreaming:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="Flow ",+ call_id="test-call-id",),)crewai_event_bus.emit(@@ -458,6 +464,7 @@ class TestFlowKickoffStreaming:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="output!",+ call_id="test-call-id",),)return "done"@@ -545,6 +552,7 @@ class TestFlowKickoffStreamingAsync:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="Async flow ",+ call_id="test-call-id",),)await asyncio.sleep(0.01)@@ -553,6 +561,7 @@ class TestFlowKickoffStreamingAsync:LLMStreamChunkEvent(type="llm_stream_chunk",chunk="stream!",+ call_id="test-call-id",),)await asyncio.sleep(0.01)@@ -686,6 +695,7 @@ class TestStreamingEdgeCases:type="llm_stream_chunk",chunk="Task 1",task_name="First task",+ call_id="test-call-id",),)return mock_outputdiff --git a/lib/crewai/tests/utilities/test_events.py b/lib/crewai/tests/utilities/test_events.pyindex 789f1f43e..81ef321d6 100644--- a/lib/crewai/tests/utilities/test_events.py+++ b/lib/crewai/tests/utilities/test_events.py@@ -984,8 +984,8 @@ def test_streaming_fallback_to_non_streaming():def mock_call(messages, tools=None, callbacks=None, available_functions=None):nonlocal fallback_called# Emit a couple of chunks to simulate partial streaming- crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="Test chunk 1", response_id = "Id"))- crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="Test chunk 2", response_id = "Id"))+ crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="Test chunk 1", response_id="Id", call_id="test-call-id"))+ crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="Test chunk 2", response_id="Id", call_id="test-call-id"))# Mark that fallback would be calledfallback_called = True@@ -1041,7 +1041,7 @@ def test_streaming_empty_response_handling():def mock_call(messages, tools=None, callbacks=None, available_functions=None):# Emit a few empty chunksfor _ in range(3):- crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="",response_id="id"))+ crewai_event_bus.emit(llm, event=LLMStreamChunkEvent(chunk="", response_id="id", call_id="test-call-id"))# Return the default message for empty responsesreturn "I apologize, but I couldn't generate a proper response. Please try again or rephrase your request."@@ -1280,6 +1280,105 @@ def test_llm_emits_event_with_lite_agent():assert set(all_agent_id) == {str(agent.id)}+# ----------- CALL_ID CORRELATION TESTS -----------+++def test_llm_call_events_share_call_id():+ """All events from a single LLM call should share the same call_id."""+ import uuid++ events = []+ condition = threading.Condition()++ @crewai_event_bus.on(LLMCallStartedEvent)+ def on_start(source, event):+ with condition:+ events.append(event)+ condition.notify()++ @crewai_event_bus.on(LLMCallCompletedEvent)+ def on_complete(source, event):+ with condition:+ events.append(event)+ condition.notify()++ llm = LLM(model="gpt-4o-mini")+ llm.call("Say hi")++ with condition:+ success = condition.wait_for(lambda: len(events) >= 2, timeout=10)+ assert success, "Timeout waiting for LLM events"++ # Behavior: all events from the call share the same call_id+ assert len(events) == 2+ assert events[0].call_id == events[1].call_id+ # call_id should be a valid UUID+ uuid.UUID(events[0].call_id)+++def test_streaming_chunks_share_call_id_with_call():+ """Streaming chunks should share call_id with started/completed events."""+ events = []+ condition = threading.Condition()++ @crewai_event_bus.on(LLMCallStartedEvent)+ def on_start(source, event):+ with condition:+ events.append(event)+ condition.notify()++ @crewai_event_bus.on(LLMStreamChunkEvent)+ def on_chunk(source, event):+ with condition:+ events.append(event)+ condition.notify()++ @crewai_event_bus.on(LLMCallCompletedEvent)+ def on_complete(source, event):+ with condition:+ events.append(event)+ condition.notify()++ llm = LLM(model="gpt-4o-mini", stream=True)+ llm.call("Say hi")++ with condition:+ # Wait for at least started, some chunks, and completed+ success = condition.wait_for(lambda: len(events) >= 3, timeout=10)+ assert success, "Timeout waiting for streaming events"++ # Behavior: all events (started, chunks, completed) share the same call_id+ call_ids = {e.call_id for e in events}+ assert len(call_ids) == 1+++def test_separate_llm_calls_have_different_call_ids():+ """Different LLM calls should have different call_ids."""+ call_ids = []+ condition = threading.Condition()++ @crewai_event_bus.on(LLMCallStartedEvent)+ def on_start(source, event):+ with condition:+ call_ids.append(event.call_id)+ condition.notify()++ llm = LLM(model="gpt-4o-mini")+ llm.call("Say hi")+ llm.call("Say bye")++ with condition:+ success = condition.wait_for(lambda: len(call_ids) >= 2, timeout=10)+ assert success, "Timeout waiting for LLM call events"++ # Behavior: each call has its own call_id+ assert len(call_ids) == 2+ assert call_ids[0] != call_ids[1]++# ----------- HUMAN FEEDBACK EVENTS -----------