refactor: improve code quality based on PR feedback

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/agents/agent_builder/base_agent.py

@@ -256,13 +256,14 @@ class BaseAgent(ABC, BaseModel):
             "tools_handler",
             "cache_handler",
             "llm",
+            "crew",  # Exclude crew to avoid circular reference
         }
 
         # Copy llm and clear callbacks
-        existing_llm = shallow_copy(self.llm)
+        existing_llm = shallow_copy(self.llm) if self.llm else None
         copied_data = self.model_dump(exclude=exclude)
         copied_data = {k: v for k, v in copied_data.items() if v is not None}
-        copied_agent = type(self)(**copied_data, llm=existing_llm, tools=self.tools)
+        copied_agent = type(self)(**copied_data, llm=existing_llm, tools=self.tools or [])
 
         return copied_agent
 

src/crewai/crew.py

@@ -1076,24 +1076,53 @@ class Crew(BaseModel):
         self,
         n_iterations: int,
         openai_model_name: Optional[str] = None,
+        llm: Optional[Union[str, LLM]] = None,
         inputs: Optional[Dict[str, Any]] = None,
     ) -> None:
-        """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures."""
+        """Test and evaluate the Crew with the given inputs for n iterations concurrently.
+        
+        Args:
+            n_iterations: Number of test iterations to run
+            openai_model_name: (Deprecated) OpenAI model name for backward compatibility
+            llm: LLM instance or model name to use for evaluation
+            inputs: Optional inputs for the crew
+        """
+        if openai_model_name:
+            warnings.warn(
+                "openai_model_name parameter is deprecated and will be removed in v3.0. Use llm parameter instead.",
+                DeprecationWarning,
+                stacklevel=2,
+            )
+        
+        if not (llm or openai_model_name):
+            raise ValueError("Either llm or openai_model_name must be provided")
+        
         test_crew = self.copy()
+        
+        # Convert string to LLM instance if needed
+        if isinstance(llm, str):
+            llm = LLM(model=llm)
+        elif openai_model_name and not llm:
+            llm = LLM(model=openai_model_name)
+        
+        assert isinstance(llm, LLM), "llm must be an LLM instance"
+        
+        try:
+            self._test_execution_span = test_crew._telemetry.test_execution_span(
+                test_crew,
+                n_iterations,
+                inputs,
+                getattr(llm, "model", None),
+            )
+            evaluator = CrewEvaluator(test_crew, llm)
 
-        self._test_execution_span = test_crew._telemetry.test_execution_span(
-            test_crew,
-            n_iterations,
-            inputs,
-            openai_model_name,  # type: ignore[arg-type]
-        )  # type: ignore[arg-type]
-        evaluator = CrewEvaluator(test_crew, openai_model_name)  # type: ignore[arg-type]
+            for i in range(1, n_iterations + 1):
+                evaluator.set_iteration(i)
+                test_crew.kickoff(inputs=inputs)
 
-        for i in range(1, n_iterations + 1):
-            evaluator.set_iteration(i)
-            test_crew.kickoff(inputs=inputs)
-
-        evaluator.print_crew_evaluation_result()
+            evaluator.print_crew_evaluation_result()
+        except Exception as e:
+            raise ValueError(f"Error during crew test execution: {str(e)}") from e
 
     def __repr__(self):
         return f"Crew(id={self.id}, process={self.process}, number_of_agents={len(self.agents)}, number_of_tasks={len(self.tasks)})"

src/crewai/utilities/evaluators/crew_evaluator_handler.py

@@ -1,4 +1,5 @@
 from collections import defaultdict
+from typing import TYPE_CHECKING, Any, Union
 
 from pydantic import BaseModel, Field
 from rich.box import HEAVY_EDGE
@@ -6,10 +7,14 @@ from rich.console import Console
 from rich.table import Table
 
 from crewai.agent import Agent
+from crewai.llm import LLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 from crewai.telemetry import Telemetry
 
+if TYPE_CHECKING:
+    from crewai.crew import Crew
+
 
 class TaskEvaluationPydanticOutput(BaseModel):
     quality: float = Field(
@@ -18,23 +23,27 @@ class TaskEvaluationPydanticOutput(BaseModel):
 
 
 class CrewEvaluator:
-    """
-    A class to evaluate the performance of the agents in the crew based on the tasks they have performed.
-
+    """Handles evaluation of Crew execution and performance.
+    
+    Args:
+        crew: The Crew instance to evaluate
+        llm: Language model to use for evaluation
+    
     Attributes:
-        crew (Crew): The crew of agents to evaluate.
-        openai_model_name (str): The model to use for evaluating the performance of the agents (for now ONLY OpenAI accepted).
-        tasks_scores (defaultdict): A dictionary to store the scores of the agents for each task.
-        iteration (int): The current iteration of the evaluation.
+        tasks_scores: Dictionary to store task scores
+        run_execution_times: Dictionary to store execution times
+        iteration: Current iteration number
+        crew: The crew instance being evaluated
+        llm: Language model used for evaluation
     """
 
-    tasks_scores: defaultdict = defaultdict(list)
-    run_execution_times: defaultdict = defaultdict(list)
+    tasks_scores: defaultdict[int, list[float]] = defaultdict(list)
+    run_execution_times: defaultdict[int, list[float]] = defaultdict(list)
     iteration: int = 0
 
-    def __init__(self, crew, openai_model_name: str):
+    def __init__(self, crew: "Crew", llm: Union[str, LLM]):
         self.crew = crew
-        self.openai_model_name = openai_model_name
+        self.llm = llm if isinstance(llm, LLM) else LLM(model=llm)
         self._telemetry = Telemetry()
         self._setup_for_evaluating()
 
@@ -51,7 +60,7 @@ class CrewEvaluator:
             ),
             backstory="Evaluator agent for crew evaluation with precise capabilities to evaluate the performance of the agents in the crew based on the tasks they have performed",
             verbose=False,
-            llm=self.openai_model_name,
+            llm=self.llm,
         )
 
     def _evaluation_task(
@@ -181,7 +190,7 @@ class CrewEvaluator:
                 self.crew,
                 evaluation_result.pydantic.quality,
                 current_task._execution_time,
-                self.openai_model_name,
+                getattr(self.llm, "model", None),
             )
             self.tasks_scores[self.iteration].append(evaluation_result.pydantic.quality)
             self.run_execution_times[self.iteration].append(

tests/crew_test.py

@@ -14,6 +14,7 @@ from crewai.agent import Agent
 from crewai.agents.cache import CacheHandler
 from crewai.crew import Crew
 from crewai.crews.crew_output import CrewOutput
+from crewai.llm import LLM
 from crewai.memory.contextual.contextual_memory import ContextualMemory
 from crewai.process import Process
 from crewai.task import Task
@@ -25,6 +26,9 @@ from crewai.utilities import Logger
 from crewai.utilities.rpm_controller import RPMController
 from crewai.utilities.task_output_storage_handler import TaskOutputStorageHandler
 
+TEST_MODEL = "gpt-4o"
+TEST_ITERATIONS = 1
+
 ceo = Agent(
     role="CEO",
     goal="Make sure the writers in your company produce amazing content.",
@@ -662,6 +666,33 @@ def test_task_tools_override_agent_tools_with_allow_delegation():
     assert isinstance(researcher_with_delegation.tools[0], TestTool)
 
 @pytest.mark.vcr(filter_headers=["authorization"])
+class TestCrewCustomLLM:
+    def test_crew_test_with_custom_llm(self):
+        tasks = [
+            Task(
+                description="Test task",
+                expected_output="Test output",
+                agent=researcher,
+            )
+        ]
+        crew = Crew(agents=[researcher], tasks=tasks)
+        
+        # Test with LLM instance
+        custom_llm = LLM(model=TEST_MODEL)
+        crew.test(n_iterations=TEST_ITERATIONS, llm=custom_llm)
+        
+        # Test with model name string
+        crew.test(n_iterations=TEST_ITERATIONS, llm=TEST_MODEL)
+        
+        # Test backward compatibility
+        crew.test(n_iterations=TEST_ITERATIONS, openai_model_name=TEST_MODEL)
+        
+        # Test error when no LLM provided
+        with pytest.raises(ValueError):
+            crew.test(n_iterations=TEST_ITERATIONS)
+
+
+
 def test_crew_verbose_output(capsys):
     tasks = [
         Task(
@@ -1123,7 +1154,7 @@ def test_kickoff_for_each_empty_input():
     assert results == []
 
 
-@pytest.mark.vcr(filter_headers=["authorization"])
+@pytest.mark.vcr(filter_headeruvs=["authorization"])
 def test_kickoff_for_each_invalid_input():
     """Tests if kickoff_for_each raises TypeError for invalid input types."""
 
@@ -2835,14 +2866,23 @@ def test_crew_testing_function(kickoff_mock, copy_mock, crew_evaluator):
         [mock.call(inputs={"topic": "AI"}), mock.call(inputs={"topic": "AI"})]
     )
 
-    crew_evaluator.assert_has_calls(
-        [
-            mock.call(crew, "gpt-4o-mini"),
-            mock.call().set_iteration(1),
-            mock.call().set_iteration(2),
-            mock.call().print_crew_evaluation_result(),
-        ]
-    )
+    # Get the actual calls made to crew_evaluator
+    actual_calls = crew_evaluator.mock_calls
+    
+    # Check that the first call was made with correct crew and either string or LLM instance
+    first_call = actual_calls[0]
+    assert first_call[0] == '', "First call should be to constructor"
+    assert first_call[1][0] == crew, "First argument should be crew"
+    assert isinstance(first_call[1][1], (str, LLM)), "Second argument should be string or LLM"
+    if isinstance(first_call[1][1], LLM):
+        assert first_call[1][1].model == "gpt-4o-mini"
+    else:
+        assert first_call[1][1] == "gpt-4o-mini"
+    
+    # Check remaining calls
+    assert actual_calls[1] == mock.call().set_iteration(1)
+    assert actual_calls[2] == mock.call().set_iteration(2)
+    assert actual_calls[3] == mock.call().print_crew_evaluation_result()
 
 
 @pytest.mark.vcr(filter_headers=["authorization"])
@@ -3125,4 +3165,4 @@ def test_multimodal_agent_live_image_analysis():
     # Verify we got a meaningful response
     assert isinstance(result.raw, str)
     assert len(result.raw) > 100  # Expecting a detailed analysis
-    assert "error" not in result.raw.lower()  # No error messages in response
+    assert "error" not in result.raw.lower()  # No error messages in response

tests/utilities/evaluators/test_crew_evaluator_handler.py

@@ -4,6 +4,7 @@ import pytest
 
 from crewai.agent import Agent
 from crewai.crew import Crew
+from crewai.llm import LLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 from crewai.utilities.evaluators.crew_evaluator_handler import (
@@ -23,7 +24,7 @@ class TestCrewEvaluator:
         )
         crew = Crew(agents=[agent], tasks=[task])
 
-        return CrewEvaluator(crew, openai_model_name="gpt-4o-mini")
+        return CrewEvaluator(crew, llm=LLM(model="gpt-4o-mini"))
 
     def test_setup_for_evaluating(self, crew_planner):
         crew_planner._setup_for_evaluating()
@@ -45,6 +46,7 @@ class TestCrewEvaluator:
             == "Evaluator agent for crew evaluation with precise capabilities to evaluate the performance of the agents in the crew based on the tasks they have performed"
         )
         assert agent.verbose is False
+        assert isinstance(agent.llm, LLM)
         assert agent.llm.model == "gpt-4o-mini"
 
     def test_evaluation_task(self, crew_planner):