Integrations / CrewAI
Monitor CrewAI agents
To monitor CrewAI with Tracify, instrument your crew with the Python SDK. Each agent step, tool call and model call is recorded in one trace for the crew run.
Your code
CrewAI
Tracify
One run.
- 01Agent and task steps
- 02Tool calls between agents
- 03Cost per crew run
- 04Failures and retries
Setup
A few lines.
python
import tracify
from openai import OpenAI
# Wrap the provider client your crew uses; calls join the current run.
openai = tracify.wrap_openai(OpenAI())Full details in the documentation. All integrations are listed on the integrations page.
- Can I see which agent in a crew caused a failure?
- Yes, each agent's steps are separate spans, so you can open the failing span directly.
