Respan BeeAI Instrumentation

Respan instrumentation plugin for the BeeAI Framework.

The package wraps openinference-instrumentation-beeai with Respan's instrumentation lifecycle and registers Respan's OpenInference translator so BeeAI agent, workflow, tool, chat model, and embedding spans are exported through the Respan OTLP pipeline.

Install

pip install respan-ai respan-instrumentation-beeai beeai-framework

Usage

Initialize Respan before creating or running BeeAI Framework components.

import asyncio

from beeai_framework.agents.requirement import RequirementAgent
from beeai_framework.backend import ChatModel
from respan import Respan
from respan_instrumentation_beeai import BeeAIInstrumentor


async def main() -> None:
    respan = Respan(instrumentations=[BeeAIInstrumentor()])

    agent = RequirementAgent(
        llm=ChatModel.from_name("openai:gpt-4.1-nano"),
        role="friendly AI assistant",
        instructions="Answer clearly and concisely.",
    )
    response = await agent.run("Explain why observability matters for agents.")
    print(response.last_message.text)

    respan.flush()


if __name__ == "__main__":
    asyncio.run(main())

Any keyword arguments passed to BeeAIInstrumentor(...) are forwarded to the underlying OpenInference BeeAI instrumentor.