OpenInference crewAI Instrumentation

Python auto-instrumentation library for LLM agents implemented with CrewAI

Crews are fully OpenTelemetry-compatible and can be sent to an OpenTelemetry collector for monitoring, such as Arize Phoenix or Arize AX.

Installation

pip install openinference-instrumentation-crewai

Quickstart

This quickstart shows you how to instrument your guardrailed LLM application

Install required packages.

pip install crewai crewai-tools  arize-phoenix opentelemetry-sdk opentelemetry-exporter-otlp

Start Phoenix in the background as a collector. By default, it listens on http://localhost:6006. You can visit the app via a browser at the same address. (Phoenix does not send data over the internet. It only operates locally on your machine.)

python -m phoenix.server.main serve

Set up CrewAIInstrumentor to trace your crew and send the traces to Phoenix at the endpoint defined below.

from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor

from openinference.instrumentation.crewai import CrewAIInstrumentor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor

endpoint = "http://127.0.0.1:6006/v1/traces"
trace_provider = TracerProvider()
trace_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter(endpoint)))

CrewAIInstrumentor().instrument(tracer_provider=trace_provider)

Set up a simple crew to do research

import os
from crewai import Agent, Task, Crew, Process
from crewai_tools import SerperDevTool

os.environ["OPENAI_API_KEY"] = "YOUR_OPENAI_API_KEY"
os.environ["SERPER_API_KEY"] = "YOUR_SERPER_API_KEY" 
search_tool = SerperDevTool()

# Define your agents with roles and goals
researcher = Agent(
  role='Senior Research Analyst',
  goal='Uncover cutting-edge developments in AI and data science',
  backstory="""You work at a leading tech think tank.
  Your expertise lies in identifying emerging trends.
  You have a knack for dissecting complex data and presenting actionable insights.""",
  verbose=True,
  allow_delegation=False,
  # You can pass an optional llm attribute specifying what model you wanna use.
  # llm=ChatOpenAI(model_name="gpt-3.5", temperature=0.7),
  tools=[search_tool]
)
writer = Agent(
  role='Tech Content Strategist',
  goal='Craft compelling content on tech advancements',
  backstory="""You are a renowned Content Strategist, known for your insightful and engaging articles.
  You transform complex concepts into compelling narratives.""",
  verbose=True,
  allow_delegation=True
)

# Create tasks for your agents
task1 = Task(
  description="""Conduct a comprehensive analysis of the latest advancements in AI in 2024.
  Identify key trends, breakthrough technologies, and potential industry impacts.""",
  expected_output="Full analysis report in bullet points",
  agent=researcher
)

task2 = Task(
  description="""Using the insights provided, develop an engaging blog
  post that highlights the most significant AI advancements.
  Your post should be informative yet accessible, catering to a tech-savvy audience.
  Make it sound cool, avoid complex words so it doesn't sound like AI.""",
  expected_output="Full blog post of at least 4 paragraphs",
  agent=writer
)

# Instantiate your crew with a sequential process
crew = Crew(
  agents=[researcher, writer],
  tasks=[task1, task2],
  verbose=True,
  process=Process.sequential
)

# Get your crew to work!
result = crew.kickoff()

print("######################")
print(result)

Event Listener Mode

CrewAIInstrumentor().instrument(...) without extra flags is the default wrapper-based integration and remains the recommended path for standard Python CrewAI applications.

Use use_event_listener=True only when CrewAI execution is surfaced through the event bus rather than direct Python method calls, such as AMP / low-code CrewAI usage. See examples/event_listener_crew.py for that setup.

By default, event-listener mode also creates LLM spans from CrewAI's LLMCall* events. That is useful when the listener is your only source of LLM visibility. If you already instrument the underlying LLM client separately, or if you want tests that focus only on crew / agent / tool structure to avoid provider- and retry-driven LLM span count variability, disable them with:

CrewAIInstrumentor().instrument(
    tracer_provider=trace_provider,
    use_event_listener=True,
    create_llm_spans=False,
)

More Info

• More info on OpenInference and Phoenix
• More info on OpenInference and Arize AX
• How to customize spans to track sessions, metadata, etc.
• How to account for private information and span payload customization