Implementation:CrewAIInc CrewAI Task Constructor

Knowledge Sources	CrewAI CrewAI Docs
Domains	Task_Management, Multi_Agent_Systems
Last Updated	2026-02-11 00:00 GMT

Overview

Concrete class for defining task specifications with descriptions, expected outputs, agent assignment, and structured output schemas provided by the CrewAI framework.

Description

The Task class is a Pydantic model representing a discrete unit of work. It requires a description (supporting {variable} interpolation) and expected_output. Optional fields include the assigned agent, output format (Pydantic model, JSON, or file), task context dependencies, tools, guardrails, human review flags, and async execution mode.

Usage

Import and instantiate Task for each work item in a crew. At minimum, provide description and expected_output. Assign an agent for sequential workflows, or leave unassigned for hierarchical manager routing.

Code Reference

Source Location

• Repository: crewAI
• File: lib/crewai/src/crewai/task.py
• Lines: L84-230

Signature

class Task(BaseModel):
    """Class that represents a task to be executed."""

    name: str | None = Field(default=None)
    description: str = Field(description="Description of the actual task.")
    expected_output: str = Field(description="Clear definition of expected output.")
    agent: BaseAgent | None = Field(default=None)
    context: list[Task] | None = Field(default=NOT_SPECIFIED)
    async_execution: bool | None = Field(default=False)
    output_json: type[BaseModel] | None = Field(default=None)
    output_pydantic: type[BaseModel] | None = Field(default=None)
    response_model: type[BaseModel] | None = Field(default=None)
    output_file: str | None = Field(default=None)
    create_directory: bool | None = Field(default=True)
    tools: list[BaseTool] | None = Field(default_factory=list)
    human_input: bool | None = Field(default=False)
    markdown: bool | None = Field(default=False)
    callback: Any | None = Field(default=None)
    guardrail: GuardrailType | None = Field(default=None)
    guardrails: GuardrailsType | None = Field(default=None)
    guardrail_max_retries: int = Field(default=3)

Import

from crewai import Task

I/O Contract

Inputs

Name	Type	Required	Description
description	str	Yes	Task description with optional {variable} interpolation
expected_output	str	Yes	Clear specification of expected output
agent	None	No	Agent assigned to execute (default: None)
output_pydantic	None	No	Pydantic model for structured output
output_json	None	No	Pydantic model for JSON output
context	None	No	Tasks whose output provides context
tools	None	No	Task-specific tools
human_input	bool	No	Require human review (default: False)

Outputs

Name	Type	Description
Task instance	Task	Configured task ready for crew execution
output (after execution)	TaskOutput	Contains raw, pydantic, json_dict, agent fields

Usage Examples

Basic Task

from crewai import Task

research_task = Task(
    description="Research the latest developments in {topic}",
    expected_output="A comprehensive summary of findings with sources",
    agent=researcher,
)

Task with Structured Output

from crewai import Task
from pydantic import BaseModel

class ResearchReport(BaseModel):
    title: str
    findings: list[str]
    recommendations: list[str]

report_task = Task(
    description="Analyze {topic} and produce a structured report",
    expected_output="A structured research report",
    agent=researcher,
    output_pydantic=ResearchReport,
)

Related Pages

Implements Principle

• Principle:CrewAIInc_CrewAI_Task_Definition
• Environment:CrewAIInc_CrewAI_Python_Runtime_Environment