Defining Agents & Tools

A good CrewAI agent is like a good employee: clear role, clear goal, the right tools. In this lesson you'll learn to define agents precisely and equip them with custom tools.

Agent Configuration in Detail

Role

The role determines how the agent behaves. Be specific:

Goal

The goal defines the agent's success metric:

analyst = Agent(
    role="Senior Data Analyst",
    goal="Create precise, data-driven analyses with clear "
         "actionable recommendations for management",
    backstory="You have 10 years of experience in data analysis for "
              "Fortune 500 companies. You communicate complex "
              "findings clearly and actionably."
)

Backstory

The backstory gives the agent context and personality. It influences communication style and decision-making.

Creating Custom Tools

Tool with the @tool Decorator

from crewai.tools import tool

@tool("Web Search")
def web_search(query: str) -> str:
    """Searches the web for current information.
    Use this tool for research on current topics."""
    results = search_api.search(query, num_results=5)
    return format_results(results)

Tool Class for Complex Logic

from crewai.tools import BaseTool
from pydantic import BaseModel, Field

class DatabaseQueryInput(BaseModel):
    query: str = Field(description="SQL query for the database")
    database: str = Field(default="analytics", description="Database name")

class DatabaseQueryTool(BaseTool):
    name: str = "Database Query"
    description: str = "Executes SQL queries on the analytics database"
    args_schema: type[BaseModel] = DatabaseQueryInput

    def _run(self, query: str, database: str = "analytics") -> str:
        return db.execute(query, database)

Tool Delegation

When allow_delegation=True, an agent can delegate tasks to other agents:

manager = Agent(
    role="Project Manager",
    goal="Coordinate the team for optimal results",
    allow_delegation=True,  # Can delegate to other agents
    tools=[project_tracker, communication_tool]
)

specialist = Agent(
    role="Domain Expert",
    goal="Deliver deep domain expertise",
    allow_delegation=False,  # Executes tasks itself
    tools=[knowledge_base, calculation_tool]
)

Memory Configuration

CrewAI offers three memory types:

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    memory=True,  # Enables long-term and entity memory
    embedder={
        "provider": "openai",
        "config": {"model": "text-embedding-3-small"}
    }
)

Practical tip: Invest time in precise roles and backstories. An agent with "Research Analyst" as its role delivers generic results. An agent with "Senior B2B SaaS Market Research Analyst focused on the DACH region" delivers relevant results.