CrewAI


Agent structure	Yes — crew, per-agent task spans, tool executions, nested.
The LLM leg	Yes, inside the tree.
Retries	Yes — CrewAI retries a failing task, and each attempt is its own span.

Agent structure

Yes — crew, per-agent task spans, tool executions, nested.

The LLM leg

Yes, inside the tree.

Retries

Yes — CrewAI retries a failing task, and each attempt is its own span.

Setup

pip install openinference-instrumentation-crewai \
            opentelemetry-sdk opentelemetry-exporter-otlp-proto-http

from opentelemetry import trace
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.resources import Resource
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from openinference.instrumentation.crewai import CrewAIInstrumentor
import os

provider = TracerProvider(resource=Resource.create({"service.name": "my-crew"}))
provider.add_span_processor(BatchSpanProcessor(OTLPSpanExporter(
    endpoint="https://gateway.tracelane.dev/v1/traces",
    headers={"x-tracelane-api-key": os.environ["TRACELANE_API_KEY"]},
)))
trace.set_tracer_provider(provider)

CrewAIInstrumentor().instrument(tracer_provider=provider)

Your API key needs the ingest scope.

Span volume — CrewAI is the heaviest of the four

A crew multiplies: agents × tasks × retries, plus a span per tool call and per LLM call. CrewAI retries a failing task, and each attempt is a separate span — a single failing task was observed emitting three. A five-agent crew can emit hundreds of spans for one kickoff().

Plan limits are counted in spans on this path.

An OTel span and a gateway call are not the same unit, and they are not counted the same way. A gateway request and an OTel span both count as ingest bytes against the SAME meter — there is no separate quota for one path or the other. On this OTel path your app sends spans, and one agent run can emit many spans. Size your ingest estimate against spans per run, not runs, and see pricing.

Verification status

The transport is the standard OTLP/HTTP protobuf exporter and is proven against Tracelane production.

The CrewAI instrumentor itself has not been run end to end against production — a realistic crew needs live LLM calls. The instrumentor attaches cleanly; the span shape above is from CrewAI’s documented instrumentation, not from a Tracelane production read-back. Treat the volume guidance as the reliable part and tell us what you see.

Get started

SDKs

Integrations

Concepts

Features

Tools

Audit & compliance

Operations

Migrations

Comparisons

What you get

Setup

Span volume — CrewAI is the heaviest of the four

Verification status

​What you get

​Setup

​Span volume — CrewAI is the heaviest of the four

​Verification status

What you get

Setup

Span volume — CrewAI is the heaviest of the four

Verification status