Implementation:CrewAIInc CrewAI Crew Kickoff

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

Overview

Concrete method for triggering crew workflow execution with input interpolation and process routing provided by the CrewAI framework.

Description

The Crew.kickoff() method is the primary entry point for executing a crew's workflow. It interpolates input variables into task descriptions, selects the execution process (sequential or hierarchical), runs the task pipeline, executes after-kickoff callbacks, and returns a CrewOutput (or CrewStreamingOutput if streaming is enabled). It also manages telemetry spans via OpenTelemetry and handles error events.

Usage

Call crew.kickoff() after assembling a Crew. Pass an optional inputs dictionary whose keys correspond to {variable} placeholders in task descriptions and expected outputs.

Code Reference

Source Location

Repository: crewAI
File: lib/crewai/src/crewai/crew.py
Lines: L696-776

Signature

def kickoff(
    self,
    inputs: dict[str, Any] | None = None,
    input_files: dict[str, FileInput] | None = None,
) -> CrewOutput | CrewStreamingOutput:
    """Execute the crew's workflow.

    Args:
        inputs: Optional input dictionary for task interpolation.
        input_files: Optional dict of named file inputs for the crew.

    Returns:
        CrewOutput or CrewStreamingOutput if streaming is enabled.
    """

Import

from crewai import Crew
# kickoff is a method on Crew instances

I/O Contract

Inputs

Name	Type	Required	Description
inputs	None	No	Input dict for {variable} interpolation in task descriptions
input_files	None	No	Named file inputs for multimodal processing

Outputs

Name	Type	Description
result	CrewOutput	Contains raw text, pydantic model, json_dict, tasks_output, token_usage
result (streaming)	CrewStreamingOutput	Streaming iterator with real-time chunks

Usage Examples

Basic Execution

result = crew.kickoff(inputs={"topic": "AI in Healthcare"})

print(result.raw)           # Raw text output
print(result.token_usage)   # Token consumption metrics

With Structured Output

result = crew.kickoff(inputs={"topic": "AI Agents"})

# If final task has output_pydantic set:
report = result.pydantic  # Typed Pydantic model instance
print(report.title)
print(report.findings)

Related Pages

Implements Principle

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment