cf.run tools cannot take in a built-in/imported functions as input

[afrizalhasbi]

Description

Running this code would give it a validation error:

import numpy as np

cf.run(
    "Convert to np array: [1,2,3]",
    result_type=str,
    agents=[agent_name],
    tools=[np.array],
)

{
	"name": "ValidationError",
	"message": "1 validation error for Task
tools
  Value error, Invalid tool: <built-in function array> [type=value_error, input_value=[<built-in function array>], input_type=list]
    For further information visit https://errors.pydantic.dev/2.9/v/value_error",
	"stack": "---------------------------------------------------------------------------
ValidationError                           Traceback (most recent call last)
Cell In[18], line 1
----> 1 cf.run(
      2     \"Convert to np array: [1,2,3]\",
      3     result_type=str,
      4     agents=[thinker],
      5     tools=[np.array],
      6 )

File ~/.local/lib/python3.12/site-packages/controlflow/run.py:109, in run(objective, turn_strategy, max_llm_calls, max_agent_turns, raise_on_failure, handlers, **task_kwargs)
     99 def run(
    100     objective: str,
    101     *,
   (...)
    107     **task_kwargs,
    108 ) -> Any:
--> 109     task = Task(objective=objective, **task_kwargs)
    110     results = run_tasks(
    111         tasks=[task],
    112         raise_on_failure=raise_on_failure,
   (...)
    116         handlers=handlers,
    117     )
    118     return results[0]

File ~/.local/lib/python3.12/site-packages/controlflow/tasks/task.py:215, in Task.__init__(self, objective, user_access, **kwargs)
    209     warnings.warn(
    210         \"The `user_access` argument is deprecated. Use `interactive=True` instead.\",
    211         DeprecationWarning,
    212     )
    213     kwargs[\"interactive\"] = True
--> 215 super().__init__(**kwargs)
    217 # create dependencies to tasks passed in as depends_on
    218 for task in self.depends_on:

File ~/.local/lib/python3.12/site-packages/pydantic/main.py:209, in BaseModel.__init__(self, **data)
    207 # `__tracebackhide__` tells pytest and some other tools to omit this function from tracebacks
    208 __tracebackhide__ = True
--> 209 validated_self = self.__pydantic_validator__.validate_python(data, self_instance=self)
    210 if self is not validated_self:
    211     warnings.warn(
    212         'A custom validator is returning a value other than `self`.\
'
    213         \"Returning anything other than `self` from a top level model validator isn't supported when validating via `__init__`.\
\"
    214         'See the `model_validator` docs (https://docs.pydantic.dev/latest/concepts/validators/#model-validators) for more details.',
    215         category=None,
    216     )

ValidationError: 1 validation error for Task
tools
  Value error, Invalid tool: <built-in function array> [type=value_error, input_value=[<built-in function array>], input_type=list]
    For further information visit https://errors.pydantic.dev/2.9/v/value_error"
}

The same is true for built-in functions, like print, and functions import with from module import function

A workaround is creating a wrapper that just does the original imported function:

import numpy as np

def getarray(input):
    return np.array(input)

cf.run(
    "Convert to np array: [1,2,3]",
    result_type=str,
    agents=[agent_name],
    tools=[getarray],
)

I was trying to add all modules from os to the run so the model can interact with my file system, but I stumbled upon this error. To do that I had to manually create 1-line functions that just wrap over the original which became tedious. Would love to see some fix/explanation to this.

Version Information

controlflow==0.10.0
pydantic==2.9.2