CrewAI

// Overview

CrewAI is a Python framework for building multi-agent systems. You give each agent a role, a goal, and a backstory, then group several of them into a "crew" that splits a job into tasks and works through them together. It is written from scratch and does not depend on LangChain or other agent frameworks.

It is aimed at Python developers who want more than a single LLM call but don't want to wire up agent orchestration by hand. CrewAI offers two main building blocks: Crews, which lean toward agent autonomy and collaboration, and Flows, an event-driven model for more precise, step-by-step control.

As a multi-agent framework, it sits between low-level LLM SDKs and full hosted platforms. You scaffold a project with the CLI, describe agents and tasks in YAML, and run the crew locally or connect it to your own model and tools.

// What it does

• Role-based agents defined by a role, goal, and backstory
• Crews that coordinate multiple agents across a set of tasks
• Flows for event-driven, step-by-step orchestration with single LLM calls
• CLI to scaffold a project (crewai create) and run it (crewai run)
• YAML configuration for agents and tasks, kept separate from Python code
• Standalone framework, independent of LangChain or other agent libraries

// Getting started

You need Python >=3.10 <3.14. Install the package, scaffold a project with the CLI, then run your crew.

uv pip install crewai
# with optional tools:
uv pip install 'crewai[tools]'

crewai create crew <project_name>

@CrewBase
class ExampleCrew():
    @agent
    def researcher(self) -> Agent:
        return Agent(config=self.agents_config['researcher'])

    @task
    def research_task(self) -> Task:
        return Task(config=self.tasks_config['research_task'])

    @crew
    def crew(self) -> Crew:
        return Crew(agents=self.agents, tasks=self.tasks)

crewai run
# or:
python src/my_project/main.py

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Splitting a research-and-write workflow across a researcher agent and a writer agent
• Automating multi-step business processes that need several specialized agents
• Building event-driven pipelines with Flows when you need precise control over each step
• Prototyping multi-agent ideas locally before connecting them to your own models and tools

// How CrewAI compares

CrewAI alongside other open-source multi-agent systems tools AI/TLDR tracks, ranked by GitHub stars.