CrewAI

AgentV provides native support for CrewAI, allowing you to evaluate role-based agent orchestration and collaborative task execution.

1. Setup Your Agent API

Wrap your CrewAI agents in a standard HTTP endpoint.

from crewai import Crew, Agent, Task
from fastapi import FastAPI

app = FastAPI()


@app.post("/execute_task")
async def execute(request: dict):
    # Map the task_description to a CrewAI Task
    # researcher = Agent(role='Researcher', ...)
    # task = Task(description=request["task_description"], agent=researcher)
    # crew = Crew(agents=[researcher], tasks=[task])
    # result = crew.kickoff()
    return {"action": "final_answer", "summary": str(result)}

2. Run Evaluation

Use the crewai protocol to connect the harness to your agent service.

agentv run \
  --path scenarios/loan_scenario.json \
  --protocol crewai \
  --agent crewai://localhost:8000/execute_task \
  --agent-name "CrewAI-Tech-Support"

3. High-Fidelity Auditing

The crewai:// protocol utilizes the Behavioral DNA bus to track task hand-offs between individual agents in the crew, providing a clear audit trail of who did what and why.

# View the agent hand-offs in the Visual Debugger
agentv console