edsl 0.1.39.dev1__py3-none-any.whl → 0.1.39.dev3__py3-none-any.whl

edsl/agents/Agent.py

@@ -1,867 +1,867 @@
-"""An Agent is an AI agent that can reference a set of traits in answering questions."""
-
-from __future__ import annotations
-import copy
-import inspect
-import types
-from typing import Callable, Optional, Union, Any, TYPE_CHECKING
-
-if TYPE_CHECKING:
-    from edsl import Cache, Survey, Scenario
-    from edsl.language_models import LanguageModel
-    from edsl.surveys.MemoryPlan import MemoryPlan
-    from edsl.questions import QuestionBase
-    from edsl.agents.Invigilator import InvigilatorBase
-
-from uuid import uuid4
-
-from edsl.Base import Base
-from edsl.prompts import Prompt
-from edsl.exceptions import QuestionScenarioRenderError
-
-from edsl.exceptions.agents import (
-    AgentErrors,
-    AgentCombinationError,
-    AgentDirectAnswerFunctionError,
-    AgentDynamicTraitsFunctionError,
-)
-
-from edsl.agents.descriptors import (
-    TraitsDescriptor,
-    CodebookDescriptor,
-    InstructionDescriptor,
-    NameDescriptor,
-)
-from edsl.utilities.decorators import (
-    sync_wrapper,
-    add_edsl_version,
-    remove_edsl_version,
-)
-from edsl.data_transfer_models import AgentResponseDict
-from edsl.utilities.restricted_python import create_restricted_function
-
-
-class Agent(Base):
-    """An class representing an agent that can answer questions."""
-
-    __doc__ = "https://docs.expectedparrot.com/en/latest/agents.html"
-
-    default_instruction = """You are answering questions as if you were a human. Do not break character."""
-
-    _traits = TraitsDescriptor()
-    codebook = CodebookDescriptor()
-    instruction = InstructionDescriptor()
-    name = NameDescriptor()
-    dynamic_traits_function_name = ""
-    answer_question_directly_function_name = ""
-    has_dynamic_traits_function = False
-
-    def __init__(
-        self,
-        traits: Optional[dict] = None,
-        name: Optional[str] = None,
-        codebook: Optional[dict] = None,
-        instruction: Optional[str] = None,
-        traits_presentation_template: Optional[str] = None,
-        dynamic_traits_function: Optional[Callable] = None,
-        dynamic_traits_function_source_code: Optional[str] = None,
-        dynamic_traits_function_name: Optional[str] = None,
-        answer_question_directly_source_code: Optional[str] = None,
-        answer_question_directly_function_name: Optional[str] = None,
-    ):
-        """Initialize a new instance of Agent.
-
-        :param traits: A dictionary of traits that the agent has. The keys need to be valid identifiers.
-        :param name: A name for the agent
-        :param codebook: A codebook mapping trait keys to trait descriptions.
-        :param instruction: Instructions for the agent in how to answer questions.
-        :param trait_presentation_template: A template for how to present the agent's traits.
-        :param dynamic_traits_function: A function that returns a dictionary of traits.
-
-        The `traits` parameter is a dictionary of traits that the agent has.
-        These traits are used to construct a prompt that is presented to the LLM.
-        In the absence of a `traits_presentation_template`, the default is used.
-        This is a template that is used to present the agent's traits to the LLM.
-        See :py:class:`edsl.prompts.library.agent_persona.AgentPersona` for more information.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.traits
-        {'age': 10, 'hair': 'brown', 'height': 5.5}
-
-        These traits are used to construct a prompt that is presented to the LLM.
-
-        In the absence of a `traits_presentation_template`, the default is used.
-
-        >>> a = Agent(traits = {"age": 10}, traits_presentation_template = "I am a {{age}} year old.")
-        >>> repr(a.agent_persona)
-        'Prompt(text=\"""I am a {{age}} year old.\""")'
-
-        When this is rendered for presentation to the LLM, it will replace the `{{age}}` with the actual age.
-        it is also possible to use the `codebook` to provide a more human-readable description of the trait.
-        Here is an example where we give a prefix to the age trait (namely the age):
-
-        >>> traits = {"age": 10, "hair": "brown", "height": 5.5}
-        >>> codebook = {'age': 'Their age is'}
-        >>> a = Agent(traits = traits, codebook = codebook, traits_presentation_template = "This agent is Dave. {{codebook['age']}} {{age}}")
-        >>> d = a.traits | {'codebook': a.codebook}
-        >>> a.agent_persona.render(d)
-        Prompt(text=\"""This agent is Dave. Their age is 10\""")
-
-        Instructions
-        ------------
-        The agent can also have instructions. These are instructions that are given to the agent when answering questions.
-
-        >>> Agent.default_instruction
-        'You are answering questions as if you were a human. Do not break character.'
-
-        See see how these are used to actually construct the prompt that is presented to the LLM, see :py:class:`edsl.agents.Invigilator.InvigilatorBase`.
-
-        """
-        self.name = name
-        self._traits = traits or dict()
-        self.codebook = codebook or dict()
-        if instruction is None:
-            self.instruction = self.default_instruction
-        else:
-            self.instruction = instruction
-        # self.instruction = instruction or self.default_instruction
-        self.dynamic_traits_function = dynamic_traits_function
-
-        # Deal with dynamic traits function
-        if self.dynamic_traits_function:
-            self.dynamic_traits_function_name = self.dynamic_traits_function.__name__
-            self.has_dynamic_traits_function = True
-        else:
-            self.has_dynamic_traits_function = False
-
-        if dynamic_traits_function_source_code:
-            self.dynamic_traits_function_name = dynamic_traits_function_name
-            self.dynamic_traits_function = create_restricted_function(
-                dynamic_traits_function_name, dynamic_traits_function
-            )
-
-        # Deal with direct answer function
-        if answer_question_directly_source_code:
-            self.answer_question_directly_function_name = (
-                answer_question_directly_function_name
-            )
-            protected_method = create_restricted_function(
-                answer_question_directly_function_name,
-                answer_question_directly_source_code,
-            )
-            bound_method = types.MethodType(protected_method, self)
-            setattr(self, "answer_question_directly", bound_method)
-
-        self._check_dynamic_traits_function()
-
-        self.current_question = None
-
-        if traits_presentation_template is not None:
-            self._traits_presentation_template = traits_presentation_template
-            self.traits_presentation_template = traits_presentation_template
-        else:
-            self.traits_presentation_template = "Your traits: {{traits}}"
-
-    @property
-    def agent_persona(self) -> Prompt:
-        return Prompt(text=self.traits_presentation_template)
-
-    def prompt(self) -> str:
-        """Return the prompt for the agent.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.prompt()
-        Prompt(text=\"""Your traits: {'age': 10, 'hair': 'brown', 'height': 5.5}\""")
-        """
-        replacement_dict = (
-            self.traits | {"traits": self.traits} | {"codebook": self.codebook}
-        )
-        if undefined := self.agent_persona.undefined_template_variables(
-            replacement_dict
-        ):
-            raise QuestionScenarioRenderError(
-                f"Agent persona still has variables that were not rendered: {undefined}"
-            )
-        else:
-            return self.agent_persona.render(replacement_dict)
-
-    def _check_dynamic_traits_function(self) -> None:
-        """Check whether dynamic trait function is valid.
-
-        This checks whether the dynamic traits function is valid.
-
-        >>> def f(question): return {"age": 10, "hair": "brown", "height": 5.5}
-        >>> a = Agent(dynamic_traits_function = f)
-        >>> a._check_dynamic_traits_function()
-
-        >>> def g(question, poo): return {"age": 10, "hair": "brown", "height": 5.5}
-        >>> a = Agent(dynamic_traits_function = g)
-        Traceback (most recent call last):
-        ...
-        edsl.exceptions.agents.AgentDynamicTraitsFunctionError: ...
-        """
-        if self.has_dynamic_traits_function:
-            sig = inspect.signature(self.dynamic_traits_function)
-            if "question" in sig.parameters:
-                if len(sig.parameters) > 1:
-                    raise AgentDynamicTraitsFunctionError(
-                        message=f"The dynamic traits function {self.dynamic_traits_function} has too many parameters. It should only have one parameter: 'question'."
-                    )
-            else:
-                if len(sig.parameters) > 0:
-                    raise AgentDynamicTraitsFunctionError(
-                        f"""The dynamic traits function {self.dynamic_traits_function} has too many parameters. It should have no parameters or 
-                        just a single parameter: 'question'."""
-                    )
-
-    @property
-    def traits(self) -> dict[str, str]:
-        """An agent's traits, which is a dictionary.
-
-        The agent could have a a dynamic traits function (`dynamic_traits_function`) that returns a dictionary of traits
-        when called. This function can also take a `question` as an argument.
-        If so, the dynamic traits function is called and the result is returned.
-        Otherwise, the traits are returned.
-
-        Example:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.traits
-        {'age': 10, 'hair': 'brown', 'height': 5.5}
-
-        """
-        if self.has_dynamic_traits_function:
-            sig = inspect.signature(self.dynamic_traits_function)
-            if "question" in sig.parameters:
-                return self.dynamic_traits_function(question=self.current_question)
-            else:
-                return self.dynamic_traits_function()
-        else:
-            return self._traits
-
-    def _repr_html_(self):
-        # d = self.to_dict(add_edsl_version=False)
-        d = self.traits
-        data = [[k, v] for k, v in d.items()]
-        from tabulate import tabulate
-
-        table = str(tabulate(data, headers=["keys", "values"], tablefmt="html"))
-        return f"<pre>{table}</pre>"
-
-    def rename(
-        self, old_name_or_dict: Union[str, dict], new_name: Optional[str] = None
-    ) -> Agent:
-        """Rename a trait.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.rename("age", "years") == Agent(traits = {'years': 10, 'hair': 'brown', 'height': 5.5})
-        True
-
-        >>> a.rename({'years': 'smage'})
-        Agent(traits = {'hair': 'brown', 'height': 5.5, 'smage': 10})
-
-        """
-        if isinstance(old_name_or_dict, dict) and new_name is None:
-            for old_name, new_name in old_name_or_dict.items():
-                self = self._rename(old_name, new_name)
-            return self
-
-        if isinstance(old_name_or_dict, dict) and new_name:
-            raise AgentErrors(
-                f"You passed a dict: {old_name_or_dict} and a new name: {new_name}. You should pass only a dict."
-            )
-
-        if isinstance(old_name_or_dict, str):
-            self._rename(old_name_or_dict, new_name)
-            return self
-
-        raise AgentErrors("Something is not right with Agent renaming")
-
-    def _rename(self, old_name: str, new_name: str) -> Agent:
-        """Rename a trait.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.rename("age", "years") == Agent(traits = {'years': 10, 'hair': 'brown', 'height': 5.5})
-        True
-        """
-        self.traits[new_name] = self.traits.pop(old_name)
-        return self
-
-    def __getitem__(self, key):
-        """Allow for accessing traits using the bracket notation.
-
-        Example:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a['traits']['age']
-        10
-
-        """
-        return getattr(self, key)
-
-    def remove_direct_question_answering_method(self) -> None:
-        """Remove the direct question answering method.
-
-        Example usage:
-
-        >>> a = Agent()
-        >>> def f(self, question, scenario): return "I am a direct answer."
-        >>> a.add_direct_question_answering_method(f)
-        >>> a.remove_direct_question_answering_method()
-        >>> hasattr(a, "answer_question_directly")
-        False
-        """
-        if hasattr(self, "answer_question_directly"):
-            delattr(self, "answer_question_directly")
-
-    def add_direct_question_answering_method(
-        self,
-        method: Callable,
-        validate_response: bool = False,
-        translate_response: bool = False,
-    ) -> None:
-        """Add a method to the agent that can answer a particular question type.
-        https://docs.expectedparrot.com/en/latest/agents.html#agent-direct-answering-methods
-
-        :param method: A method that can answer a question directly.
-        :param validate_response: Whether to validate the response.
-        :param translate_response: Whether to translate the response.
-
-        Example usage:
-
-        >>> a = Agent()
-        >>> def f(self, question, scenario): return "I am a direct answer."
-        >>> a.add_direct_question_answering_method(f)
-        >>> a.answer_question_directly(question = None, scenario = None)
-        'I am a direct answer.'
-        """
-        if hasattr(self, "answer_question_directly"):
-            import warnings
-
-            warnings.warn(
-                "Warning: overwriting existing answer_question_directly method"
-            )
-
-        self.validate_response = validate_response
-        self.translate_response = translate_response
-
-        signature = inspect.signature(method)
-        for argument in ["question", "scenario", "self"]:
-            if argument not in signature.parameters:
-                raise AgentDirectAnswerFunctionError(
-                    f"The method {method} does not have a '{argument}' parameter."
-                )
-        bound_method = types.MethodType(method, self)
-        setattr(self, "answer_question_directly", bound_method)
-        self.answer_question_directly_function_name = bound_method.__name__
-
-    def create_invigilator(
-        self,
-        *,
-        question: "QuestionBase",
-        cache: "Cache",
-        survey: Optional["Survey"] = None,
-        scenario: Optional["Scenario"] = None,
-        model: Optional["LanguageModel"] = None,
-        debug: bool = False,
-        memory_plan: Optional["MemoryPlan"] = None,
-        current_answers: Optional[dict] = None,
-        iteration: int = 1,
-        sidecar_model=None,
-        raise_validation_errors: bool = True,
-    ) -> "InvigilatorBase":
-        """Create an Invigilator.
-
-        An invigilator is an object that is responsible for administering a question to an agent.
-        There are several different types of invigilators, depending on the type of question and the agent.
-        For example, there are invigilators for functional questions (i.e., question is of type :class:`edsl.questions.QuestionFunctional`:), for direct questions, and for LLM questions.
-
-        >>> a = Agent(traits = {})
-        >>> a.create_invigilator(question = None, cache = False)
-        InvigilatorAI(...)
-
-        An invigator is an object that is responsible for administering a question to an agent and
-        recording the responses.
-        """
-        from edsl import Model, Scenario
-
-        cache = cache
-        self.current_question = question
-        model = model or Model()
-        scenario = scenario or Scenario()
-        invigilator = self._create_invigilator(
-            question=question,
-            scenario=scenario,
-            survey=survey,
-            model=model,
-            debug=debug,
-            memory_plan=memory_plan,
-            current_answers=current_answers,
-            iteration=iteration,
-            cache=cache,
-            sidecar_model=sidecar_model,
-            raise_validation_errors=raise_validation_errors,
-        )
-        if hasattr(self, "validate_response"):
-            invigilator.validate_response = self.validate_response
-        if hasattr(self, "translate_response"):
-            invigilator.translate_response = self.translate_response
-        return invigilator
-
-    async def async_answer_question(
-        self,
-        *,
-        question: QuestionBase,
-        cache: Cache,
-        scenario: Optional[Scenario] = None,
-        survey: Optional[Survey] = None,
-        model: Optional[LanguageModel] = None,
-        debug: bool = False,
-        memory_plan: Optional[MemoryPlan] = None,
-        current_answers: Optional[dict] = None,
-        iteration: int = 0,
-    ) -> AgentResponseDict:
-        """
-        Answer a posed question.
-
-        :param question: The question to answer.
-        :param scenario: The scenario in which the question is asked.
-        :param model: The language model to use.
-        :param debug: Whether to run in debug mode.
-        :param memory_plan: The memory plan to use.
-        :param current_answers: The current answers.
-        :param iteration: The iteration number.
-
-        >>> a = Agent(traits = {})
-        >>> a.add_direct_question_answering_method(lambda self, question, scenario: "I am a direct answer.")
-        >>> from edsl import QuestionFreeText
-        >>> q = QuestionFreeText.example()
-        >>> a.answer_question(question = q, cache = False).answer
-        'I am a direct answer.'
-
-        This is a function where an agent returns an answer to a particular question.
-        However, there are several different ways an agent can answer a question, so the
-        actual functionality is delegated to an :class:`edsl.agents.InvigilatorBase`: object.
-        """
-        invigilator = self.create_invigilator(
-            question=question,
-            cache=cache,
-            scenario=scenario,
-            survey=survey,
-            model=model,
-            debug=debug,
-            memory_plan=memory_plan,
-            current_answers=current_answers,
-            iteration=iteration,
-        )
-        response: AgentResponseDict = await invigilator.async_answer_question()
-        return response
-
-    answer_question = sync_wrapper(async_answer_question)
-
-    def _create_invigilator(
-        self,
-        question: QuestionBase,
-        cache: Optional[Cache] = None,
-        scenario: Optional[Scenario] = None,
-        model: Optional[LanguageModel] = None,
-        survey: Optional[Survey] = None,
-        debug: bool = False,
-        memory_plan: Optional[MemoryPlan] = None,
-        current_answers: Optional[dict] = None,
-        iteration: int = 0,
-        sidecar_model=None,
-        raise_validation_errors: bool = True,
-    ) -> "InvigilatorBase":
-        """Create an Invigilator."""
-        from edsl import Model
-        from edsl import Scenario
-
-        model = model or Model()
-        scenario = scenario or Scenario()
-
-        from edsl.agents.Invigilator import (
-            InvigilatorHuman,
-            InvigilatorFunctional,
-            InvigilatorAI,
-            InvigilatorBase,
-        )
-
-        if cache is None:
-            from edsl.data.Cache import Cache
-
-            cache = Cache()
-
-        if debug:
-            raise NotImplementedError("Debug mode is not yet implemented.")
-            # use the question's _simulate_answer method
-            # invigilator_class = InvigilatorDebug
-        elif hasattr(question, "answer_question_directly"):
-            # It's a functional question and the answer only depends on the agent's traits & the scenario
-            invigilator_class = InvigilatorFunctional
-        elif hasattr(self, "answer_question_directly"):
-            # this of the case where the agent has a method that can answer the question directly
-            # this occurrs when 'answer_question_directly' has been given to the
-            # which happens when the agent is created from an existing survey
-            invigilator_class = InvigilatorHuman
-        else:
-            # this means an LLM agent will be used. This is the standard case.
-            invigilator_class = InvigilatorAI
-
-        if sidecar_model is not None:
-            # this is the case when a 'simple' model is being used
-            from edsl.agents.Invigilator import InvigilatorSidecar
-
-            invigilator_class = InvigilatorSidecar
-
-        invigilator = invigilator_class(
-            self,
-            question=question,
-            scenario=scenario,
-            survey=survey,
-            model=model,
-            memory_plan=memory_plan,
-            current_answers=current_answers,
-            iteration=iteration,
-            cache=cache,
-            sidecar_model=sidecar_model,
-            raise_validation_errors=raise_validation_errors,
-        )
-        return invigilator
-
-    def select(self, *traits: str) -> Agent:
-        """Selects agents with only the references traits
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-
-
-        >>> a.select("age", "height")
-        Agent(traits = {'age': 10, 'height': 5.5})
-
-        >>> a.select("age")
-        Agent(traits = {'age': 10})
-
-        """
-
-        if len(traits) == 1:
-            traits_to_select = [list(traits)[0]]
-        else:
-            traits_to_select = list(traits)
-
-        return Agent(traits={trait: self.traits[trait] for trait in traits_to_select})
-
-    def __add__(self, other_agent: Optional[Agent] = None) -> Agent:
-        """
-        Combine two agents by joining their traits.
-
-        The agents must not have overlapping traits.
-
-        Example usage:
-
-        >>> a1 = Agent(traits = {"age": 10})
-        >>> a2 = Agent(traits = {"height": 5.5})
-        >>> a1 + a2
-        Agent(traits = {'age': 10, 'height': 5.5})
-        >>> a1 + a1
-        Traceback (most recent call last):
-        ...
-        edsl.exceptions.agents.AgentCombinationError: The agents have overlapping traits: {'age'}.
-        ...
-        """
-        if other_agent is None:
-            return self
-        elif common_traits := set(self.traits.keys()) & set(other_agent.traits.keys()):
-            raise AgentCombinationError(
-                f"The agents have overlapping traits: {common_traits}."
-            )
-        else:
-            new_agent = Agent(traits=copy.deepcopy(self.traits))
-            new_agent.traits.update(other_agent.traits)
-            return new_agent
-
-    def __eq__(self, other: Agent) -> bool:
-        """Check if two agents are equal.
-
-        This only checks the traits.
-        >>> a1 = Agent(traits = {"age": 10})
-        >>> a2 = Agent(traits = {"age": 10})
-        >>> a1 == a2
-        True
-        >>> a3 = Agent(traits = {"age": 11})
-        >>> a1 == a3
-        False
-        """
-        return self.data == other.data
-
-    def __getattr__(self, name):
-        # This will be called only if 'name' is not found in the usual places
-        if name == "has_dynamic_traits_function":
-            return self.has_dynamic_traits_function
-
-        if name in self._traits:
-            return self._traits[name]
-
-        raise AttributeError(
-            f"'{type(self).__name__}' object has no attribute '{name}'"
-        )
-
-    def __getstate__(self):
-        state = self.__dict__.copy()
-        # Include any additional state that needs to be serialized
-        return state
-
-    def __setstate__(self, state):
-        self.__dict__.update(state)
-        # Ensure _traits is initialized if it's missing
-        if "_traits" not in self.__dict__:
-            self._traits = {}
-
-    def print(self) -> None:
-        from rich import print_json
-        import json
-
-        print_json(json.dumps(self.to_dict()))
-
-    def __repr__(self) -> str:
-        """Return representation of Agent."""
-        class_name = self.__class__.__name__
-        items = [
-            f'{k} = """{v}"""' if isinstance(v, str) else f"{k} = {v}"
-            for k, v in self.data.items()
-            if k != "question_type"
-        ]
-        return f"{class_name}({', '.join(items)})"
-
-    # def _repr_html_(self):
-    #     from edsl.utilities.utilities import data_to_html
-
-    #     return data_to_html(self.to_dict())
-
-    #######################
-    # SERIALIZATION METHODS
-    #######################
-    @property
-    def data(self) -> dict:
-        """Format the data for serialization.
-
-        TODO: Warn if has dynamic traits function or direct answer function that cannot be serialized.
-        TODO: Add ability to have coop-hosted functions that are serializable.
-        """
-
-        raw_data = {
-            k.replace("_", "", 1): v
-            for k, v in self.__dict__.items()
-            if k.startswith("_")
-        }
-
-        if hasattr(self, "set_instructions"):
-            if not self.set_instructions:
-                raw_data.pop("instruction")
-        if self.codebook == {}:
-            raw_data.pop("codebook")
-        if self.name == None:
-            raw_data.pop("name")
-
-        if hasattr(self, "dynamic_traits_function"):
-            raw_data.pop(
-                "dynamic_traits_function", None
-            )  # in case dynamic_traits_function will appear with _ in self.__dict__
-            dynamic_traits_func = self.dynamic_traits_function
-            if dynamic_traits_func:
-                func = inspect.getsource(dynamic_traits_func)
-                raw_data["dynamic_traits_function_source_code"] = func
-                raw_data[
-                    "dynamic_traits_function_name"
-                ] = self.dynamic_traits_function_name
-        if hasattr(self, "answer_question_directly"):
-            raw_data.pop(
-                "answer_question_directly", None
-            )  # in case answer_question_directly will appear with _ in self.__dict__
-            answer_question_directly_func = self.answer_question_directly
-
-            if (
-                answer_question_directly_func
-                and raw_data.get("answer_question_directly_source_code", None) != None
-            ):
-                raw_data["answer_question_directly_source_code"] = inspect.getsource(
-                    answer_question_directly_func
-                )
-                raw_data[
-                    "answer_question_directly_function_name"
-                ] = self.answer_question_directly_function_name
-
-        return raw_data
-
-    def __hash__(self) -> int:
-        from edsl.utilities.utilities import dict_hash
-
-        return dict_hash(self.to_dict(add_edsl_version=False))
-
-    # @add_edsl_version
-    def to_dict(self, add_edsl_version=True) -> dict[str, Union[dict, bool]]:
-        """Serialize to a dictionary with EDSL info.
-
-        Example usage:
-
-        >>> a = Agent(name = "Steve", traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.to_dict()
-        {'name': 'Steve', 'traits': {'age': 10, 'hair': 'brown', 'height': 5.5}, 'edsl_version': '...', 'edsl_class_name': 'Agent'}
-        """
-        d = copy.deepcopy(self.data)
-        if add_edsl_version:
-            from edsl import __version__
-
-            d["edsl_version"] = __version__
-            d["edsl_class_name"] = self.__class__.__name__
-
-        return d
-
-    @classmethod
-    @remove_edsl_version
-    def from_dict(cls, agent_dict: dict[str, Union[dict, bool]]) -> Agent:
-        """Deserialize from a dictionary.
-
-        Example usage:
-
-        >>> Agent.from_dict({'name': "Steve", 'traits': {'age': 10, 'hair': 'brown', 'height': 5.5}})
-        Agent(name = \"""Steve\""", traits = {'age': 10, 'hair': 'brown', 'height': 5.5})
-
-        """
-        return cls(**agent_dict)
-
-    def _table(self) -> tuple[dict, list]:
-        """Prepare generic table data."""
-        table_data = []
-        for attr_name, attr_value in self.__dict__.items():
-            table_data.append({"Attribute": attr_name, "Value": repr(attr_value)})
-        column_names = ["Attribute", "Value"]
-        return table_data, column_names
-
-    def add_trait(self, trait_name_or_dict: str, value: Optional[Any] = None) -> Agent:
-        """Adds a trait to an agent and returns that agent"""
-        if isinstance(trait_name_or_dict, dict) and value is None:
-            self.traits.update(trait_name_or_dict)
-            return self
-
-        if isinstance(trait_name_or_dict, dict) and value:
-            raise AgentErrors(
-                f"You passed a dict: {trait_name_or_dict} and a value: {value}. You should pass only a dict."
-            )
-
-        if isinstance(trait_name_or_dict, str):
-            trait = trait_name_or_dict
-            self.traits[trait] = value
-            return self
-
-        raise AgentErrors("Something is not right with adding a trait to an Agent")
-
-    def remove_trait(self, trait: str) -> Agent:
-        """Remove a trait from the agent.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.remove_trait("age")
-        Agent(traits = {'hair': 'brown', 'height': 5.5})
-        """
-        _ = self.traits.pop(trait)
-        return self
-
-    def translate_traits(self, values_codebook: dict) -> Agent:
-        """Translate traits to a new codebook.
-
-        >>> a = Agent(traits = {"age": 10, "hair": 1, "height": 5.5})
-        >>> a.translate_traits({"hair": {1:"brown"}})
-        Agent(traits = {'age': 10, 'hair': 'brown', 'height': 5.5})
-
-        :param values_codebook: The new codebook.
-        """
-        for key, value in self.traits.items():
-            if key in values_codebook:
-                self.traits[key] = values_codebook[key][value]
-        return self
-
-    def rich_print(self):
-        """Display an object as a rich table.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> a.rich_print()
-        <rich.table.Table object at ...>
-        """
-        from rich.table import Table
-
-        table_data, column_names = self._table()
-        table = Table(title=f"{self.__class__.__name__} Attributes")
-        for column in column_names:
-            table.add_column(column, style="bold")
-
-        for row in table_data:
-            row_data = [row[column] for column in column_names]
-            table.add_row(*row_data)
-
-        return table
-
-    @classmethod
-    def example(cls, randomize: bool = False) -> Agent:
-        """
-        Returns an example Agent instance.
-
-        :param randomize: If True, adds a random string to the value of an example key.
-        """
-        addition = "" if not randomize else str(uuid4())
-        return cls(traits={"age": 22, "hair": f"brown{addition}", "height": 5.5})
-
-    def code(self) -> str:
-        """Return the code for the agent.
-
-        Example usage:
-
-        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
-        >>> print(a.code())
-        from edsl import Agent
-        agent = Agent(traits={'age': 10, 'hair': 'brown', 'height': 5.5})
-        """
-        return f"from edsl import Agent\nagent = Agent(traits={self.traits})"
-
-
-def main():
-    """
-    Give an example of usage.
-
-    WARNING: Consume API credits
-    """
-    from edsl.agents import Agent
-    from edsl.questions import QuestionMultipleChoice
-
-    # a simple agent
-    agent = Agent(traits={"age": 10, "hair": "brown", "height": 5.5})
-    agent.traits
-    agent.print()
-    # combining two agents
-    agent = Agent(traits={"age": 10}) + Agent(traits={"height": 5.5})
-    agent.traits
-    # Agent -> Job using the to() method
-    agent = Agent(traits={"allergies": "peanut"})
-    question = QuestionMultipleChoice(
-        question_text="Would you enjoy a PB&J?",
-        question_options=["Yes", "No"],
-        question_name="food_preference",
-    )
-    job = question.by(agent)
-    results = job.run()
-
-
-if __name__ == "__main__":
-    import doctest
-
-    doctest.testmod(optionflags=doctest.ELLIPSIS)
+"""An Agent is an AI agent that can reference a set of traits in answering questions."""
+
+from __future__ import annotations
+import copy
+import inspect
+import types
+from typing import Callable, Optional, Union, Any, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from edsl import Cache, Survey, Scenario
+    from edsl.language_models import LanguageModel
+    from edsl.surveys.MemoryPlan import MemoryPlan
+    from edsl.questions import QuestionBase
+    from edsl.agents.Invigilator import InvigilatorBase
+
+from uuid import uuid4
+
+from edsl.Base import Base
+from edsl.prompts import Prompt
+from edsl.exceptions import QuestionScenarioRenderError
+
+from edsl.exceptions.agents import (
+    AgentErrors,
+    AgentCombinationError,
+    AgentDirectAnswerFunctionError,
+    AgentDynamicTraitsFunctionError,
+)
+
+from edsl.agents.descriptors import (
+    TraitsDescriptor,
+    CodebookDescriptor,
+    InstructionDescriptor,
+    NameDescriptor,
+)
+from edsl.utilities.decorators import (
+    sync_wrapper,
+    add_edsl_version,
+    remove_edsl_version,
+)
+from edsl.data_transfer_models import AgentResponseDict
+from edsl.utilities.restricted_python import create_restricted_function
+
+
+class Agent(Base):
+    """An class representing an agent that can answer questions."""
+
+    __doc__ = "https://docs.expectedparrot.com/en/latest/agents.html"
+
+    default_instruction = """You are answering questions as if you were a human. Do not break character."""
+
+    _traits = TraitsDescriptor()
+    codebook = CodebookDescriptor()
+    instruction = InstructionDescriptor()
+    name = NameDescriptor()
+    dynamic_traits_function_name = ""
+    answer_question_directly_function_name = ""
+    has_dynamic_traits_function = False
+
+    def __init__(
+        self,
+        traits: Optional[dict] = None,
+        name: Optional[str] = None,
+        codebook: Optional[dict] = None,
+        instruction: Optional[str] = None,
+        traits_presentation_template: Optional[str] = None,
+        dynamic_traits_function: Optional[Callable] = None,
+        dynamic_traits_function_source_code: Optional[str] = None,
+        dynamic_traits_function_name: Optional[str] = None,
+        answer_question_directly_source_code: Optional[str] = None,
+        answer_question_directly_function_name: Optional[str] = None,
+    ):
+        """Initialize a new instance of Agent.
+
+        :param traits: A dictionary of traits that the agent has. The keys need to be valid identifiers.
+        :param name: A name for the agent
+        :param codebook: A codebook mapping trait keys to trait descriptions.
+        :param instruction: Instructions for the agent in how to answer questions.
+        :param trait_presentation_template: A template for how to present the agent's traits.
+        :param dynamic_traits_function: A function that returns a dictionary of traits.
+
+        The `traits` parameter is a dictionary of traits that the agent has.
+        These traits are used to construct a prompt that is presented to the LLM.
+        In the absence of a `traits_presentation_template`, the default is used.
+        This is a template that is used to present the agent's traits to the LLM.
+        See :py:class:`edsl.prompts.library.agent_persona.AgentPersona` for more information.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.traits
+        {'age': 10, 'hair': 'brown', 'height': 5.5}
+
+        These traits are used to construct a prompt that is presented to the LLM.
+
+        In the absence of a `traits_presentation_template`, the default is used.
+
+        >>> a = Agent(traits = {"age": 10}, traits_presentation_template = "I am a {{age}} year old.")
+        >>> repr(a.agent_persona)
+        'Prompt(text=\"""I am a {{age}} year old.\""")'
+
+        When this is rendered for presentation to the LLM, it will replace the `{{age}}` with the actual age.
+        it is also possible to use the `codebook` to provide a more human-readable description of the trait.
+        Here is an example where we give a prefix to the age trait (namely the age):
+
+        >>> traits = {"age": 10, "hair": "brown", "height": 5.5}
+        >>> codebook = {'age': 'Their age is'}
+        >>> a = Agent(traits = traits, codebook = codebook, traits_presentation_template = "This agent is Dave. {{codebook['age']}} {{age}}")
+        >>> d = a.traits | {'codebook': a.codebook}
+        >>> a.agent_persona.render(d)
+        Prompt(text=\"""This agent is Dave. Their age is 10\""")
+
+        Instructions
+        ------------
+        The agent can also have instructions. These are instructions that are given to the agent when answering questions.
+
+        >>> Agent.default_instruction
+        'You are answering questions as if you were a human. Do not break character.'
+
+        See see how these are used to actually construct the prompt that is presented to the LLM, see :py:class:`edsl.agents.Invigilator.InvigilatorBase`.
+
+        """
+        self.name = name
+        self._traits = traits or dict()
+        self.codebook = codebook or dict()
+        if instruction is None:
+            self.instruction = self.default_instruction
+        else:
+            self.instruction = instruction
+        # self.instruction = instruction or self.default_instruction
+        self.dynamic_traits_function = dynamic_traits_function
+
+        # Deal with dynamic traits function
+        if self.dynamic_traits_function:
+            self.dynamic_traits_function_name = self.dynamic_traits_function.__name__
+            self.has_dynamic_traits_function = True
+        else:
+            self.has_dynamic_traits_function = False
+
+        if dynamic_traits_function_source_code:
+            self.dynamic_traits_function_name = dynamic_traits_function_name
+            self.dynamic_traits_function = create_restricted_function(
+                dynamic_traits_function_name, dynamic_traits_function
+            )
+
+        # Deal with direct answer function
+        if answer_question_directly_source_code:
+            self.answer_question_directly_function_name = (
+                answer_question_directly_function_name
+            )
+            protected_method = create_restricted_function(
+                answer_question_directly_function_name,
+                answer_question_directly_source_code,
+            )
+            bound_method = types.MethodType(protected_method, self)
+            setattr(self, "answer_question_directly", bound_method)
+
+        self._check_dynamic_traits_function()
+
+        self.current_question = None
+
+        if traits_presentation_template is not None:
+            self._traits_presentation_template = traits_presentation_template
+            self.traits_presentation_template = traits_presentation_template
+        else:
+            self.traits_presentation_template = "Your traits: {{traits}}"
+
+    @property
+    def agent_persona(self) -> Prompt:
+        return Prompt(text=self.traits_presentation_template)
+
+    def prompt(self) -> str:
+        """Return the prompt for the agent.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.prompt()
+        Prompt(text=\"""Your traits: {'age': 10, 'hair': 'brown', 'height': 5.5}\""")
+        """
+        replacement_dict = (
+            self.traits | {"traits": self.traits} | {"codebook": self.codebook}
+        )
+        if undefined := self.agent_persona.undefined_template_variables(
+            replacement_dict
+        ):
+            raise QuestionScenarioRenderError(
+                f"Agent persona still has variables that were not rendered: {undefined}"
+            )
+        else:
+            return self.agent_persona.render(replacement_dict)
+
+    def _check_dynamic_traits_function(self) -> None:
+        """Check whether dynamic trait function is valid.
+
+        This checks whether the dynamic traits function is valid.
+
+        >>> def f(question): return {"age": 10, "hair": "brown", "height": 5.5}
+        >>> a = Agent(dynamic_traits_function = f)
+        >>> a._check_dynamic_traits_function()
+
+        >>> def g(question, poo): return {"age": 10, "hair": "brown", "height": 5.5}
+        >>> a = Agent(dynamic_traits_function = g)
+        Traceback (most recent call last):
+        ...
+        edsl.exceptions.agents.AgentDynamicTraitsFunctionError: ...
+        """
+        if self.has_dynamic_traits_function:
+            sig = inspect.signature(self.dynamic_traits_function)
+            if "question" in sig.parameters:
+                if len(sig.parameters) > 1:
+                    raise AgentDynamicTraitsFunctionError(
+                        message=f"The dynamic traits function {self.dynamic_traits_function} has too many parameters. It should only have one parameter: 'question'."
+                    )
+            else:
+                if len(sig.parameters) > 0:
+                    raise AgentDynamicTraitsFunctionError(
+                        f"""The dynamic traits function {self.dynamic_traits_function} has too many parameters. It should have no parameters or 
+                        just a single parameter: 'question'."""
+                    )
+
+    @property
+    def traits(self) -> dict[str, str]:
+        """An agent's traits, which is a dictionary.
+
+        The agent could have a a dynamic traits function (`dynamic_traits_function`) that returns a dictionary of traits
+        when called. This function can also take a `question` as an argument.
+        If so, the dynamic traits function is called and the result is returned.
+        Otherwise, the traits are returned.
+
+        Example:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.traits
+        {'age': 10, 'hair': 'brown', 'height': 5.5}
+
+        """
+        if self.has_dynamic_traits_function:
+            sig = inspect.signature(self.dynamic_traits_function)
+            if "question" in sig.parameters:
+                return self.dynamic_traits_function(question=self.current_question)
+            else:
+                return self.dynamic_traits_function()
+        else:
+            return self._traits
+
+    def _repr_html_(self):
+        # d = self.to_dict(add_edsl_version=False)
+        d = self.traits
+        data = [[k, v] for k, v in d.items()]
+        from tabulate import tabulate
+
+        table = str(tabulate(data, headers=["keys", "values"], tablefmt="html"))
+        return f"<pre>{table}</pre>"
+
+    def rename(
+        self, old_name_or_dict: Union[str, dict], new_name: Optional[str] = None
+    ) -> Agent:
+        """Rename a trait.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.rename("age", "years") == Agent(traits = {'years': 10, 'hair': 'brown', 'height': 5.5})
+        True
+
+        >>> a.rename({'years': 'smage'})
+        Agent(traits = {'hair': 'brown', 'height': 5.5, 'smage': 10})
+
+        """
+        if isinstance(old_name_or_dict, dict) and new_name is None:
+            for old_name, new_name in old_name_or_dict.items():
+                self = self._rename(old_name, new_name)
+            return self
+
+        if isinstance(old_name_or_dict, dict) and new_name:
+            raise AgentErrors(
+                f"You passed a dict: {old_name_or_dict} and a new name: {new_name}. You should pass only a dict."
+            )
+
+        if isinstance(old_name_or_dict, str):
+            self._rename(old_name_or_dict, new_name)
+            return self
+
+        raise AgentErrors("Something is not right with Agent renaming")
+
+    def _rename(self, old_name: str, new_name: str) -> Agent:
+        """Rename a trait.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.rename("age", "years") == Agent(traits = {'years': 10, 'hair': 'brown', 'height': 5.5})
+        True
+        """
+        self.traits[new_name] = self.traits.pop(old_name)
+        return self
+
+    def __getitem__(self, key):
+        """Allow for accessing traits using the bracket notation.
+
+        Example:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a['traits']['age']
+        10
+
+        """
+        return getattr(self, key)
+
+    def remove_direct_question_answering_method(self) -> None:
+        """Remove the direct question answering method.
+
+        Example usage:
+
+        >>> a = Agent()
+        >>> def f(self, question, scenario): return "I am a direct answer."
+        >>> a.add_direct_question_answering_method(f)
+        >>> a.remove_direct_question_answering_method()
+        >>> hasattr(a, "answer_question_directly")
+        False
+        """
+        if hasattr(self, "answer_question_directly"):
+            delattr(self, "answer_question_directly")
+
+    def add_direct_question_answering_method(
+        self,
+        method: Callable,
+        validate_response: bool = False,
+        translate_response: bool = False,
+    ) -> None:
+        """Add a method to the agent that can answer a particular question type.
+        https://docs.expectedparrot.com/en/latest/agents.html#agent-direct-answering-methods
+
+        :param method: A method that can answer a question directly.
+        :param validate_response: Whether to validate the response.
+        :param translate_response: Whether to translate the response.
+
+        Example usage:
+
+        >>> a = Agent()
+        >>> def f(self, question, scenario): return "I am a direct answer."
+        >>> a.add_direct_question_answering_method(f)
+        >>> a.answer_question_directly(question = None, scenario = None)
+        'I am a direct answer.'
+        """
+        if hasattr(self, "answer_question_directly"):
+            import warnings
+
+            warnings.warn(
+                "Warning: overwriting existing answer_question_directly method"
+            )
+
+        self.validate_response = validate_response
+        self.translate_response = translate_response
+
+        signature = inspect.signature(method)
+        for argument in ["question", "scenario", "self"]:
+            if argument not in signature.parameters:
+                raise AgentDirectAnswerFunctionError(
+                    f"The method {method} does not have a '{argument}' parameter."
+                )
+        bound_method = types.MethodType(method, self)
+        setattr(self, "answer_question_directly", bound_method)
+        self.answer_question_directly_function_name = bound_method.__name__
+
+    def create_invigilator(
+        self,
+        *,
+        question: "QuestionBase",
+        cache: "Cache",
+        survey: Optional["Survey"] = None,
+        scenario: Optional["Scenario"] = None,
+        model: Optional["LanguageModel"] = None,
+        debug: bool = False,
+        memory_plan: Optional["MemoryPlan"] = None,
+        current_answers: Optional[dict] = None,
+        iteration: int = 1,
+        sidecar_model=None,
+        raise_validation_errors: bool = True,
+    ) -> "InvigilatorBase":
+        """Create an Invigilator.
+
+        An invigilator is an object that is responsible for administering a question to an agent.
+        There are several different types of invigilators, depending on the type of question and the agent.
+        For example, there are invigilators for functional questions (i.e., question is of type :class:`edsl.questions.QuestionFunctional`:), for direct questions, and for LLM questions.
+
+        >>> a = Agent(traits = {})
+        >>> a.create_invigilator(question = None, cache = False)
+        InvigilatorAI(...)
+
+        An invigator is an object that is responsible for administering a question to an agent and
+        recording the responses.
+        """
+        from edsl import Model, Scenario
+
+        cache = cache
+        self.current_question = question
+        model = model or Model()
+        scenario = scenario or Scenario()
+        invigilator = self._create_invigilator(
+            question=question,
+            scenario=scenario,
+            survey=survey,
+            model=model,
+            debug=debug,
+            memory_plan=memory_plan,
+            current_answers=current_answers,
+            iteration=iteration,
+            cache=cache,
+            sidecar_model=sidecar_model,
+            raise_validation_errors=raise_validation_errors,
+        )
+        if hasattr(self, "validate_response"):
+            invigilator.validate_response = self.validate_response
+        if hasattr(self, "translate_response"):
+            invigilator.translate_response = self.translate_response
+        return invigilator
+
+    async def async_answer_question(
+        self,
+        *,
+        question: QuestionBase,
+        cache: Cache,
+        scenario: Optional[Scenario] = None,
+        survey: Optional[Survey] = None,
+        model: Optional[LanguageModel] = None,
+        debug: bool = False,
+        memory_plan: Optional[MemoryPlan] = None,
+        current_answers: Optional[dict] = None,
+        iteration: int = 0,
+    ) -> AgentResponseDict:
+        """
+        Answer a posed question.
+
+        :param question: The question to answer.
+        :param scenario: The scenario in which the question is asked.
+        :param model: The language model to use.
+        :param debug: Whether to run in debug mode.
+        :param memory_plan: The memory plan to use.
+        :param current_answers: The current answers.
+        :param iteration: The iteration number.
+
+        >>> a = Agent(traits = {})
+        >>> a.add_direct_question_answering_method(lambda self, question, scenario: "I am a direct answer.")
+        >>> from edsl import QuestionFreeText
+        >>> q = QuestionFreeText.example()
+        >>> a.answer_question(question = q, cache = False).answer
+        'I am a direct answer.'
+
+        This is a function where an agent returns an answer to a particular question.
+        However, there are several different ways an agent can answer a question, so the
+        actual functionality is delegated to an :class:`edsl.agents.InvigilatorBase`: object.
+        """
+        invigilator = self.create_invigilator(
+            question=question,
+            cache=cache,
+            scenario=scenario,
+            survey=survey,
+            model=model,
+            debug=debug,
+            memory_plan=memory_plan,
+            current_answers=current_answers,
+            iteration=iteration,
+        )
+        response: AgentResponseDict = await invigilator.async_answer_question()
+        return response
+
+    answer_question = sync_wrapper(async_answer_question)
+
+    def _create_invigilator(
+        self,
+        question: QuestionBase,
+        cache: Optional[Cache] = None,
+        scenario: Optional[Scenario] = None,
+        model: Optional[LanguageModel] = None,
+        survey: Optional[Survey] = None,
+        debug: bool = False,
+        memory_plan: Optional[MemoryPlan] = None,
+        current_answers: Optional[dict] = None,
+        iteration: int = 0,
+        sidecar_model=None,
+        raise_validation_errors: bool = True,
+    ) -> "InvigilatorBase":
+        """Create an Invigilator."""
+        from edsl import Model
+        from edsl import Scenario
+
+        model = model or Model()
+        scenario = scenario or Scenario()
+
+        from edsl.agents.Invigilator import (
+            InvigilatorHuman,
+            InvigilatorFunctional,
+            InvigilatorAI,
+            InvigilatorBase,
+        )
+
+        if cache is None:
+            from edsl.data.Cache import Cache
+
+            cache = Cache()
+
+        if debug:
+            raise NotImplementedError("Debug mode is not yet implemented.")
+            # use the question's _simulate_answer method
+            # invigilator_class = InvigilatorDebug
+        elif hasattr(question, "answer_question_directly"):
+            # It's a functional question and the answer only depends on the agent's traits & the scenario
+            invigilator_class = InvigilatorFunctional
+        elif hasattr(self, "answer_question_directly"):
+            # this of the case where the agent has a method that can answer the question directly
+            # this occurrs when 'answer_question_directly' has been given to the
+            # which happens when the agent is created from an existing survey
+            invigilator_class = InvigilatorHuman
+        else:
+            # this means an LLM agent will be used. This is the standard case.
+            invigilator_class = InvigilatorAI
+
+        if sidecar_model is not None:
+            # this is the case when a 'simple' model is being used
+            from edsl.agents.Invigilator import InvigilatorSidecar
+
+            invigilator_class = InvigilatorSidecar
+
+        invigilator = invigilator_class(
+            self,
+            question=question,
+            scenario=scenario,
+            survey=survey,
+            model=model,
+            memory_plan=memory_plan,
+            current_answers=current_answers,
+            iteration=iteration,
+            cache=cache,
+            sidecar_model=sidecar_model,
+            raise_validation_errors=raise_validation_errors,
+        )
+        return invigilator
+
+    def select(self, *traits: str) -> Agent:
+        """Selects agents with only the references traits
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+
+
+        >>> a.select("age", "height")
+        Agent(traits = {'age': 10, 'height': 5.5})
+
+        >>> a.select("age")
+        Agent(traits = {'age': 10})
+
+        """
+
+        if len(traits) == 1:
+            traits_to_select = [list(traits)[0]]
+        else:
+            traits_to_select = list(traits)
+
+        return Agent(traits={trait: self.traits[trait] for trait in traits_to_select})
+
+    def __add__(self, other_agent: Optional[Agent] = None) -> Agent:
+        """
+        Combine two agents by joining their traits.
+
+        The agents must not have overlapping traits.
+
+        Example usage:
+
+        >>> a1 = Agent(traits = {"age": 10})
+        >>> a2 = Agent(traits = {"height": 5.5})
+        >>> a1 + a2
+        Agent(traits = {'age': 10, 'height': 5.5})
+        >>> a1 + a1
+        Traceback (most recent call last):
+        ...
+        edsl.exceptions.agents.AgentCombinationError: The agents have overlapping traits: {'age'}.
+        ...
+        """
+        if other_agent is None:
+            return self
+        elif common_traits := set(self.traits.keys()) & set(other_agent.traits.keys()):
+            raise AgentCombinationError(
+                f"The agents have overlapping traits: {common_traits}."
+            )
+        else:
+            new_agent = Agent(traits=copy.deepcopy(self.traits))
+            new_agent.traits.update(other_agent.traits)
+            return new_agent
+
+    def __eq__(self, other: Agent) -> bool:
+        """Check if two agents are equal.
+
+        This only checks the traits.
+        >>> a1 = Agent(traits = {"age": 10})
+        >>> a2 = Agent(traits = {"age": 10})
+        >>> a1 == a2
+        True
+        >>> a3 = Agent(traits = {"age": 11})
+        >>> a1 == a3
+        False
+        """
+        return self.data == other.data
+
+    def __getattr__(self, name):
+        # This will be called only if 'name' is not found in the usual places
+        if name == "has_dynamic_traits_function":
+            return self.has_dynamic_traits_function
+
+        if name in self._traits:
+            return self._traits[name]
+
+        raise AttributeError(
+            f"'{type(self).__name__}' object has no attribute '{name}'"
+        )
+
+    def __getstate__(self):
+        state = self.__dict__.copy()
+        # Include any additional state that needs to be serialized
+        return state
+
+    def __setstate__(self, state):
+        self.__dict__.update(state)
+        # Ensure _traits is initialized if it's missing
+        if "_traits" not in self.__dict__:
+            self._traits = {}
+
+    def print(self) -> None:
+        from rich import print_json
+        import json
+
+        print_json(json.dumps(self.to_dict()))
+
+    def __repr__(self) -> str:
+        """Return representation of Agent."""
+        class_name = self.__class__.__name__
+        items = [
+            f'{k} = """{v}"""' if isinstance(v, str) else f"{k} = {v}"
+            for k, v in self.data.items()
+            if k != "question_type"
+        ]
+        return f"{class_name}({', '.join(items)})"
+
+    # def _repr_html_(self):
+    #     from edsl.utilities.utilities import data_to_html
+
+    #     return data_to_html(self.to_dict())
+
+    #######################
+    # SERIALIZATION METHODS
+    #######################
+    @property
+    def data(self) -> dict:
+        """Format the data for serialization.
+
+        TODO: Warn if has dynamic traits function or direct answer function that cannot be serialized.
+        TODO: Add ability to have coop-hosted functions that are serializable.
+        """
+
+        raw_data = {
+            k.replace("_", "", 1): v
+            for k, v in self.__dict__.items()
+            if k.startswith("_")
+        }
+
+        if hasattr(self, "set_instructions"):
+            if not self.set_instructions:
+                raw_data.pop("instruction")
+        if self.codebook == {}:
+            raw_data.pop("codebook")
+        if self.name == None:
+            raw_data.pop("name")
+
+        if hasattr(self, "dynamic_traits_function"):
+            raw_data.pop(
+                "dynamic_traits_function", None
+            )  # in case dynamic_traits_function will appear with _ in self.__dict__
+            dynamic_traits_func = self.dynamic_traits_function
+            if dynamic_traits_func:
+                func = inspect.getsource(dynamic_traits_func)
+                raw_data["dynamic_traits_function_source_code"] = func
+                raw_data[
+                    "dynamic_traits_function_name"
+                ] = self.dynamic_traits_function_name
+        if hasattr(self, "answer_question_directly"):
+            raw_data.pop(
+                "answer_question_directly", None
+            )  # in case answer_question_directly will appear with _ in self.__dict__
+            answer_question_directly_func = self.answer_question_directly
+
+            if (
+                answer_question_directly_func
+                and raw_data.get("answer_question_directly_source_code", None) != None
+            ):
+                raw_data["answer_question_directly_source_code"] = inspect.getsource(
+                    answer_question_directly_func
+                )
+                raw_data[
+                    "answer_question_directly_function_name"
+                ] = self.answer_question_directly_function_name
+
+        return raw_data
+
+    def __hash__(self) -> int:
+        from edsl.utilities.utilities import dict_hash
+
+        return dict_hash(self.to_dict(add_edsl_version=False))
+
+    # @add_edsl_version
+    def to_dict(self, add_edsl_version=True) -> dict[str, Union[dict, bool]]:
+        """Serialize to a dictionary with EDSL info.
+
+        Example usage:
+
+        >>> a = Agent(name = "Steve", traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.to_dict()
+        {'name': 'Steve', 'traits': {'age': 10, 'hair': 'brown', 'height': 5.5}, 'edsl_version': '...', 'edsl_class_name': 'Agent'}
+        """
+        d = copy.deepcopy(self.data)
+        if add_edsl_version:
+            from edsl import __version__
+
+            d["edsl_version"] = __version__
+            d["edsl_class_name"] = self.__class__.__name__
+
+        return d
+
+    @classmethod
+    @remove_edsl_version
+    def from_dict(cls, agent_dict: dict[str, Union[dict, bool]]) -> Agent:
+        """Deserialize from a dictionary.
+
+        Example usage:
+
+        >>> Agent.from_dict({'name': "Steve", 'traits': {'age': 10, 'hair': 'brown', 'height': 5.5}})
+        Agent(name = \"""Steve\""", traits = {'age': 10, 'hair': 'brown', 'height': 5.5})
+
+        """
+        return cls(**agent_dict)
+
+    def _table(self) -> tuple[dict, list]:
+        """Prepare generic table data."""
+        table_data = []
+        for attr_name, attr_value in self.__dict__.items():
+            table_data.append({"Attribute": attr_name, "Value": repr(attr_value)})
+        column_names = ["Attribute", "Value"]
+        return table_data, column_names
+
+    def add_trait(self, trait_name_or_dict: str, value: Optional[Any] = None) -> Agent:
+        """Adds a trait to an agent and returns that agent"""
+        if isinstance(trait_name_or_dict, dict) and value is None:
+            self.traits.update(trait_name_or_dict)
+            return self
+
+        if isinstance(trait_name_or_dict, dict) and value:
+            raise AgentErrors(
+                f"You passed a dict: {trait_name_or_dict} and a value: {value}. You should pass only a dict."
+            )
+
+        if isinstance(trait_name_or_dict, str):
+            trait = trait_name_or_dict
+            self.traits[trait] = value
+            return self
+
+        raise AgentErrors("Something is not right with adding a trait to an Agent")
+
+    def remove_trait(self, trait: str) -> Agent:
+        """Remove a trait from the agent.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.remove_trait("age")
+        Agent(traits = {'hair': 'brown', 'height': 5.5})
+        """
+        _ = self.traits.pop(trait)
+        return self
+
+    def translate_traits(self, values_codebook: dict) -> Agent:
+        """Translate traits to a new codebook.
+
+        >>> a = Agent(traits = {"age": 10, "hair": 1, "height": 5.5})
+        >>> a.translate_traits({"hair": {1:"brown"}})
+        Agent(traits = {'age': 10, 'hair': 'brown', 'height': 5.5})
+
+        :param values_codebook: The new codebook.
+        """
+        for key, value in self.traits.items():
+            if key in values_codebook:
+                self.traits[key] = values_codebook[key][value]
+        return self
+
+    def rich_print(self):
+        """Display an object as a rich table.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> a.rich_print()
+        <rich.table.Table object at ...>
+        """
+        from rich.table import Table
+
+        table_data, column_names = self._table()
+        table = Table(title=f"{self.__class__.__name__} Attributes")
+        for column in column_names:
+            table.add_column(column, style="bold")
+
+        for row in table_data:
+            row_data = [row[column] for column in column_names]
+            table.add_row(*row_data)
+
+        return table
+
+    @classmethod
+    def example(cls, randomize: bool = False) -> Agent:
+        """
+        Returns an example Agent instance.
+
+        :param randomize: If True, adds a random string to the value of an example key.
+        """
+        addition = "" if not randomize else str(uuid4())
+        return cls(traits={"age": 22, "hair": f"brown{addition}", "height": 5.5})
+
+    def code(self) -> str:
+        """Return the code for the agent.
+
+        Example usage:
+
+        >>> a = Agent(traits = {"age": 10, "hair": "brown", "height": 5.5})
+        >>> print(a.code())
+        from edsl import Agent
+        agent = Agent(traits={'age': 10, 'hair': 'brown', 'height': 5.5})
+        """
+        return f"from edsl import Agent\nagent = Agent(traits={self.traits})"
+
+
+def main():
+    """
+    Give an example of usage.
+
+    WARNING: Consume API credits
+    """
+    from edsl.agents import Agent
+    from edsl.questions import QuestionMultipleChoice
+
+    # a simple agent
+    agent = Agent(traits={"age": 10, "hair": "brown", "height": 5.5})
+    agent.traits
+    agent.print()
+    # combining two agents
+    agent = Agent(traits={"age": 10}) + Agent(traits={"height": 5.5})
+    agent.traits
+    # Agent -> Job using the to() method
+    agent = Agent(traits={"allergies": "peanut"})
+    question = QuestionMultipleChoice(
+        question_text="Would you enjoy a PB&J?",
+        question_options=["Yes", "No"],
+        question_name="food_preference",
+    )
+    job = question.by(agent)
+    results = job.run()
+
+
+if __name__ == "__main__":
+    import doctest
+
+    doctest.testmod(optionflags=doctest.ELLIPSIS)