OpenInference Instrumentation for BeeAI

This module provides automatic instrumentation for BeeAI framework. It integrates seamlessly with the @opentelemetry/sdk-trace-node to collect and export telemetry data.

Installation

pip install openinference-instrumentation-beeai

Quickstart

This quickstart shows you how to instrument your guardrailed LLM application

Install required packages.

pip install beeai-framework 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 BeeAIInstrumentor to trace your crew and send the traces to Phoenix at the endpoint defined below. The openinference_setup.py file.

import logging

from opentelemetry import trace as trace_api
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk import trace as trace_sdk
from opentelemetry.sdk.resources import Resource
from opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor

from openinference.instrumentation.beeai import BeeAIInstrumentor

logging.basicConfig(level=logging.DEBUG)


def setup_observability(endpoint: str = "http://localhost:6006/v1/traces") -> None:
    """
    Sets up OpenTelemetry with OTLP HTTP exporter and instruments the beeai framework.
    """
    resource = Resource(attributes={})
    tracer_provider = trace_sdk.TracerProvider(resource=resource)
    tracer_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter(endpoint)))
    tracer_provider.add_span_processor(SimpleSpanProcessor(ConsoleSpanExporter()))
    trace_api.set_tracer_provider(tracer_provider)

    BeeAIInstrumentor().instrument()

Set up a simple ReActAgent to get the current weather in Las Vegas.

import asyncio
import sys
import traceback

from beeai_framework.agents.react import ReActAgent
from beeai_framework.agents.types import AgentExecutionConfig
from beeai_framework.backend.chat import ChatModel
from beeai_framework.backend.types import ChatModelParameters
from beeai_framework.errors import FrameworkError
from beeai_framework.memory import TokenMemory
from beeai_framework.tools.search import DuckDuckGoSearchTool, WikipediaTool
from beeai_framework.tools.tool import AnyTool
from beeai_framework.tools.weather.openmeteo import OpenMeteoTool
from openinference_setup import setup_observability

setup_observability()

llm = ChatModel.from_name(
    "ollama:granite3.1-dense:8b",
    ChatModelParameters(temperature=0),
)

tools: list[AnyTool] = [
    WikipediaTool(),
    OpenMeteoTool(),
    DuckDuckGoSearchTool(),
]

agent = ReActAgent(llm=llm, tools=tools, memory=TokenMemory(llm))

prompt = "What's the current weather in Las Vegas?"


async def main() -> None:
    response = await agent.run(
        prompt=prompt,
        execution=AgentExecutionConfig(
            max_retries_per_step=3, total_max_retries=10, max_iterations=20
        ),
    )

    print("Agent 🤖 : ", response.result.text)


if __name__ == "__main__":
    try:
        asyncio.run(main())
    except FrameworkError as e:
        traceback.print_exc()
        sys.exit(e.explain())

More Info

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