crewAI/src/crewai/lite_agent.py

import asyncio
import inspect
import uuid
from typing import (
    Any,
    Callable,
    Dict,
    List,
    Optional,
    Tuple,
    Type,
    Union,
    cast,
    get_args,
    get_origin,
)

try:
    from typing import Self
except ImportError:
    from typing_extensions import Self

from pydantic import (
    BaseModel,
    Field,
    InstanceOf,
    PrivateAttr,
    model_validator,
    field_validator,
)

from crewai.agents.agent_builder.base_agent import BaseAgent
from crewai.agents.agent_builder.utilities.base_token_process import TokenProcess
from crewai.agents.cache import CacheHandler
from crewai.agents.parser import (
    AgentAction,
    AgentFinish,
    OutputParserException,
)
from crewai.flow.flow_trackable import FlowTrackable
from crewai.llm import LLM
from crewai.tools.base_tool import BaseTool
from crewai.tools.structured_tool import CrewStructuredTool
from crewai.utilities import I18N
from crewai.utilities.guardrail import process_guardrail
from crewai.utilities.agent_utils import (
    enforce_rpm_limit,
    format_message_for_llm,
    get_llm_response,
    get_tool_names,
    handle_agent_action_core,
    handle_context_length,
    handle_max_iterations_exceeded,
    handle_output_parser_exception,
    handle_unknown_error,
    has_reached_max_iterations,
    is_context_length_exceeded,
    parse_tools,
    process_llm_response,
    render_text_description_and_args,
)
from crewai.utilities.converter import generate_model_description
from crewai.utilities.events.agent_events import (
    AgentLogsExecutionEvent,
    LiteAgentExecutionCompletedEvent,
    LiteAgentExecutionErrorEvent,
    LiteAgentExecutionStartedEvent,
)
from crewai.utilities.events.crewai_event_bus import crewai_event_bus
from crewai.utilities.events.llm_events import (
    LLMCallCompletedEvent,
    LLMCallFailedEvent,
    LLMCallStartedEvent,
    LLMCallType,
)
from crewai.utilities.llm_utils import create_llm
from crewai.utilities.printer import Printer
from crewai.utilities.token_counter_callback import TokenCalcHandler
from crewai.utilities.tool_utils import execute_tool_and_check_finality


class LiteAgentOutput(BaseModel):
    """Class that represents the result of a LiteAgent execution."""

    model_config = {"arbitrary_types_allowed": True}

    raw: str = Field(description="Raw output of the agent", default="")
    pydantic: Optional[BaseModel] = Field(
        description="Pydantic output of the agent", default=None
    )
    agent_role: str = Field(description="Role of the agent that produced this output")
    usage_metrics: Optional[Dict[str, Any]] = Field(
        description="Token usage metrics for this execution", default=None
    )

    def to_dict(self) -> Dict[str, Any]:
        """Convert pydantic_output to a dictionary."""
        if self.pydantic:
            return self.pydantic.model_dump()
        return {}

    def __str__(self) -> str:
        """String representation of the output."""
        if self.pydantic:
            return str(self.pydantic)
        return self.raw


class LiteAgent(FlowTrackable, BaseModel):
    """
    A lightweight agent that can process messages and use tools.

    This agent is simpler than the full Agent class, focusing on direct execution
    rather than task delegation. It's designed to be used for simple interactions
    where a full crew is not needed.

    Attributes:
        role: The role of the agent.
        goal: The objective of the agent.
        backstory: The backstory of the agent.
        llm: The language model that will run the agent.
        tools: Tools at the agent's disposal.
        verbose: Whether the agent execution should be in verbose mode.
        max_iterations: Maximum number of iterations for tool usage.
        max_execution_time: Maximum execution time in seconds.
        response_format: Optional Pydantic model for structured output.
    """

    model_config = {"arbitrary_types_allowed": True}

    # Core Agent Properties
    role: str = Field(description="Role of the agent")
    goal: str = Field(description="Goal of the agent")
    backstory: str = Field(description="Backstory of the agent")
    llm: Optional[Union[str, InstanceOf[LLM], Any]] = Field(
        default=None, description="Language model that will run the agent"
    )
    tools: List[BaseTool] = Field(
        default_factory=list, description="Tools at agent's disposal"
    )

    # Execution Control Properties
    max_iterations: int = Field(
        default=15, description="Maximum number of iterations for tool usage"
    )
    max_execution_time: Optional[int] = Field(
        default=None, description="Maximum execution time in seconds"
    )
    respect_context_window: bool = Field(
        default=True,
        description="Whether to respect the context window of the LLM",
    )
    use_stop_words: bool = Field(
        default=True,
        description="Whether to use stop words to prevent the LLM from using tools",
    )
    request_within_rpm_limit: Optional[Callable[[], bool]] = Field(
        default=None,
        description="Callback to check if the request is within the RPM limit",
    )
    i18n: I18N = Field(default=I18N(), description="Internationalization settings.")

    # Output and Formatting Properties
    response_format: Optional[Type[BaseModel]] = Field(
        default=None, description="Pydantic model for structured output"
    )
    verbose: bool = Field(
        default=False, description="Whether to print execution details"
    )
    callbacks: List[Callable] = Field(
        default=[], description="Callbacks to be used for the agent"
    )

    # Guardrail Properties
    guardrail: Optional[Union[Callable[[LiteAgentOutput], Tuple[bool, Any]], str]] = (
        Field(
            default=None,
            description="Function or string description of a guardrail to validate agent output",
        )
    )
    guardrail_max_retries: int = Field(
        default=3, description="Maximum number of retries when guardrail fails"
    )

    # State and Results
    tools_results: List[Dict[str, Any]] = Field(
        default=[], description="Results of the tools used by the agent."
    )

    # Reference of Agent
    original_agent: Optional[BaseAgent] = Field(
        default=None, description="Reference to the agent that created this LiteAgent"
    )
    # Private Attributes
    _parsed_tools: List[CrewStructuredTool] = PrivateAttr(default_factory=list)
    _token_process: TokenProcess = PrivateAttr(default_factory=TokenProcess)
    _cache_handler: CacheHandler = PrivateAttr(default_factory=CacheHandler)
    _key: str = PrivateAttr(default_factory=lambda: str(uuid.uuid4()))
    _messages: List[Dict[str, str]] = PrivateAttr(default_factory=list)
    _iterations: int = PrivateAttr(default=0)
    _printer: Printer = PrivateAttr(default_factory=Printer)
    _guardrail: Optional[Callable] = PrivateAttr(default=None)
    _guardrail_retry_count: int = PrivateAttr(default=0)

    @model_validator(mode="after")
    def setup_llm(self):
        """Set up the LLM and other components after initialization."""
        self.llm = create_llm(self.llm)
        if not isinstance(self.llm, LLM):
            raise ValueError("Unable to create LLM instance")

        # Initialize callbacks
        token_callback = TokenCalcHandler(token_cost_process=self._token_process)
        self._callbacks = [token_callback]

        return self

    @model_validator(mode="after")
    def parse_tools(self):
        """Parse the tools and convert them to CrewStructuredTool instances."""
        self._parsed_tools = parse_tools(self.tools)

        return self

    @model_validator(mode="after")
    def ensure_guardrail_is_callable(self) -> Self:
        if callable(self.guardrail):
            self._guardrail = self.guardrail
        elif isinstance(self.guardrail, str):
            from crewai.tasks.llm_guardrail import LLMGuardrail

            assert isinstance(self.llm, LLM)

            self._guardrail = LLMGuardrail(description=self.guardrail, llm=self.llm)

        return self

    @field_validator("guardrail", mode="before")
    @classmethod
    def validate_guardrail_function(
        cls, v: Optional[Union[Callable, str]]
    ) -> Optional[Union[Callable, str]]:
        """Validate that the guardrail function has the correct signature.

        If v is a callable, validate that it has the correct signature.
        If v is a string, return it as is.

        Args:
            v: The guardrail function to validate or a string describing the guardrail task

        Returns:
            The validated guardrail function or a string describing the guardrail task
        """
        if v is None or isinstance(v, str):
            return v

        # Check function signature
        sig = inspect.signature(v)
        if len(sig.parameters) != 1:
            raise ValueError(
                f"Guardrail function must accept exactly 1 parameter (LiteAgentOutput), "
                f"but it accepts {len(sig.parameters)}"
            )

        # Check return annotation if present
        if sig.return_annotation is not sig.empty:
            if sig.return_annotation == Tuple[bool, Any]:
                return v

            origin = get_origin(sig.return_annotation)
            args = get_args(sig.return_annotation)

            if origin is not tuple or len(args) != 2 or args[0] is not bool:
                raise ValueError(
                    "If return type is annotated, it must be Tuple[bool, Any]"
                )

        return v

    @property
    def key(self) -> str:
        """Get the unique key for this agent instance."""
        return self._key

    @property
    def _original_role(self) -> str:
        """Return the original role for compatibility with tool interfaces."""
        return self.role

    def kickoff(self, messages: Union[str, List[Dict[str, str]]]) -> LiteAgentOutput:
        """
        Execute the agent with the given messages.

        Args:
            messages: Either a string query or a list of message dictionaries.
                     If a string is provided, it will be converted to a user message.
                     If a list is provided, each dict should have 'role' and 'content' keys.

        Returns:
            LiteAgentOutput: The result of the agent execution.
        """
        # Create agent info for event emission
        agent_info = {
            "role": self.role,
            "goal": self.goal,
            "backstory": self.backstory,
            "tools": self._parsed_tools,
            "verbose": self.verbose,
        }

        try:
            # Reset state for this run
            self._iterations = 0
            self.tools_results = []

            # Format messages for the LLM
            self._messages = self._format_messages(messages)

            return self._execute_core(agent_info=agent_info)

        except Exception as e:
            self._printer.print(
                content="Agent failed to reach a final answer. This is likely a bug - please report it.",
                color="red",
            )
            handle_unknown_error(self._printer, e)
            # Emit error event
            crewai_event_bus.emit(
                self,
                event=LiteAgentExecutionErrorEvent(
                    agent_info=agent_info,
                    error=str(e),
                ),
            )
            raise e

    def _execute_core(self, agent_info: Dict[str, Any]) -> LiteAgentOutput:
        # Emit event for agent execution start
        crewai_event_bus.emit(
            self,
            event=LiteAgentExecutionStartedEvent(
                agent_info=agent_info,
                tools=self._parsed_tools,
                messages=self._messages,
            ),
        )

        # Execute the agent using invoke loop
        agent_finish = self._invoke_loop()
        formatted_result: Optional[BaseModel] = None
        if self.response_format:
            try:
                # Cast to BaseModel to ensure type safety
                result = self.response_format.model_validate_json(agent_finish.output)
                if isinstance(result, BaseModel):
                    formatted_result = result
            except Exception as e:
                self._printer.print(
                    content=f"Failed to parse output into response format: {str(e)}",
                    color="yellow",
                )

        # Calculate token usage metrics
        usage_metrics = self._token_process.get_summary()

        # Create output
        output = LiteAgentOutput(
            raw=agent_finish.output,
            pydantic=formatted_result,
            agent_role=self.role,
            usage_metrics=usage_metrics.model_dump() if usage_metrics else None,
        )

        # Process guardrail if set
        if self._guardrail is not None:
            guardrail_result = process_guardrail(
                output=output,
                guardrail=self._guardrail,
                retry_count=self._guardrail_retry_count,
            )

            if not guardrail_result.success:
                if self._guardrail_retry_count >= self.guardrail_max_retries:
                    raise Exception(
                        f"Agent's guardrail failed validation after {self.guardrail_max_retries} retries. "
                        f"Last error: {guardrail_result.error}"
                    )
                self._guardrail_retry_count += 1
                if self.verbose:
                    self._printer.print(
                        f"Guardrail failed. Retrying ({self._guardrail_retry_count}/{self.guardrail_max_retries})..."
                        f"\n{guardrail_result.error}"
                    )

                self._messages.append(
                    {
                        "role": "user",
                        "content": guardrail_result.error
                        or "Guardrail validation failed",
                    }
                )

                return self._execute_core(agent_info=agent_info)

            # Apply guardrail result if available
            if guardrail_result.result is not None:
                if isinstance(guardrail_result.result, str):
                    output.raw = guardrail_result.result
                elif isinstance(guardrail_result.result, BaseModel):
                    output.pydantic = guardrail_result.result

            usage_metrics = self._token_process.get_summary()
            output.usage_metrics = usage_metrics.model_dump() if usage_metrics else None

        # Emit completion event
        crewai_event_bus.emit(
            self,
            event=LiteAgentExecutionCompletedEvent(
                agent_info=agent_info,
                output=agent_finish.output,
            ),
        )

        return output

    async def kickoff_async(
        self, messages: Union[str, List[Dict[str, str]]]
    ) -> LiteAgentOutput:
        """
        Execute the agent asynchronously with the given messages.

        Args:
            messages: Either a string query or a list of message dictionaries.
                     If a string is provided, it will be converted to a user message.
                     If a list is provided, each dict should have 'role' and 'content' keys.

        Returns:
            LiteAgentOutput: The result of the agent execution.
        """
        return await asyncio.to_thread(self.kickoff, messages)

    def _get_default_system_prompt(self) -> str:
        """Get the default system prompt for the agent."""
        base_prompt = ""
        if self._parsed_tools:
            # Use the prompt template for agents with tools
            base_prompt = self.i18n.slice("lite_agent_system_prompt_with_tools").format(
                role=self.role,
                backstory=self.backstory,
                goal=self.goal,
                tools=render_text_description_and_args(self._parsed_tools),
                tool_names=get_tool_names(self._parsed_tools),
            )
        else:
            # Use the prompt template for agents without tools
            base_prompt = self.i18n.slice(
                "lite_agent_system_prompt_without_tools"
            ).format(
                role=self.role,
                backstory=self.backstory,
                goal=self.goal,
            )

        # Add response format instructions if specified
        if self.response_format:
            schema = generate_model_description(self.response_format)
            base_prompt += self.i18n.slice("lite_agent_response_format").format(
                response_format=schema
            )

        return base_prompt

    def _format_messages(
        self, messages: Union[str, List[Dict[str, str]]]
    ) -> List[Dict[str, str]]:
        """Format messages for the LLM."""
        if isinstance(messages, str):
            messages = [{"role": "user", "content": messages}]

        system_prompt = self._get_default_system_prompt()

        # Add system message at the beginning
        formatted_messages = [{"role": "system", "content": system_prompt}]

        # Add the rest of the messages
        formatted_messages.extend(messages)

        return formatted_messages

    def _invoke_loop(self) -> AgentFinish:
        """
        Run the agent's thought process until it reaches a conclusion or max iterations.

        Returns:
            AgentFinish: The final result of the agent execution.
        """
        # Execute the agent loop
        formatted_answer = None
        while not isinstance(formatted_answer, AgentFinish):
            try:
                if has_reached_max_iterations(self._iterations, self.max_iterations):
                    formatted_answer = handle_max_iterations_exceeded(
                        formatted_answer,
                        printer=self._printer,
                        i18n=self.i18n,
                        messages=self._messages,
                        llm=cast(LLM, self.llm),
                        callbacks=self._callbacks,
                    )

                enforce_rpm_limit(self.request_within_rpm_limit)

                # Emit LLM call started event
                crewai_event_bus.emit(
                    self,
                    event=LLMCallStartedEvent(
                        messages=self._messages,
                        tools=None,
                        callbacks=self._callbacks,
                    ),
                )

                try:
                    answer = get_llm_response(
                        llm=cast(LLM, self.llm),
                        messages=self._messages,
                        callbacks=self._callbacks,
                        printer=self._printer,
                    )

                    # Emit LLM call completed event
                    crewai_event_bus.emit(
                        self,
                        event=LLMCallCompletedEvent(
                            response=answer,
                            call_type=LLMCallType.LLM_CALL,
                        ),
                    )
                except Exception as e:
                    # Emit LLM call failed event
                    crewai_event_bus.emit(
                        self,
                        event=LLMCallFailedEvent(error=str(e)),
                    )
                    raise e

                formatted_answer = process_llm_response(answer, self.use_stop_words)

                if isinstance(formatted_answer, AgentAction):
                    try:
                        tool_result = execute_tool_and_check_finality(
                            agent_action=formatted_answer,
                            tools=self._parsed_tools,
                            i18n=self.i18n,
                            agent_key=self.key,
                            agent_role=self.role,
                            agent=self.original_agent,
                        )
                    except Exception as e:
                        raise e

                    formatted_answer = handle_agent_action_core(
                        formatted_answer=formatted_answer,
                        tool_result=tool_result,
                        show_logs=self._show_logs,
                    )

                self._append_message(formatted_answer.text, role="assistant")
            except OutputParserException as e:
                formatted_answer = handle_output_parser_exception(
                    e=e,
                    messages=self._messages,
                    iterations=self._iterations,
                    log_error_after=3,
                    printer=self._printer,
                )

            except Exception as e:
                if e.__class__.__module__.startswith("litellm"):
                    # Do not retry on litellm errors
                    raise e
                if is_context_length_exceeded(e):
                    handle_context_length(
                        respect_context_window=self.respect_context_window,
                        printer=self._printer,
                        messages=self._messages,
                        llm=cast(LLM, self.llm),
                        callbacks=self._callbacks,
                        i18n=self.i18n,
                    )
                    continue
                else:
                    handle_unknown_error(self._printer, e)
                    raise e

            finally:
                self._iterations += 1

        assert isinstance(formatted_answer, AgentFinish)
        self._show_logs(formatted_answer)
        return formatted_answer

    def _show_logs(self, formatted_answer: Union[AgentAction, AgentFinish]):
        """Show logs for the agent's execution."""
        crewai_event_bus.emit(
            self,
            AgentLogsExecutionEvent(
                agent_role=self.role,
                formatted_answer=formatted_answer,
                verbose=self.verbose,
            ),
        )

    def _append_message(self, text: str, role: str = "assistant") -> None:
        """Append a message to the message list with the given role."""
        self._messages.append(format_message_for_llm(text, role=role))