Respan LiveKit Instrumentation

Respan instrumentation plugin for LiveKit Agents. It translates LiveKit's native livekit-agents LLM spans into canonical Respan chat spans and emits Respan tool spans for execute_function_call.

Installation

pip install respan-ai respan-instrumentation-livekit livekit-agents

Usage

import os

from livekit.agents import llm
from respan import Respan, workflow
from respan_instrumentation_livekit import LiveKitInstrumentor

respan = Respan(
    api_key=os.environ["RESPAN_API_KEY"],
    app_name="livekit-example",
    instrumentations=[LiveKitInstrumentor()],
)


@workflow(name="livekit_chat.workflow")
async def run_chat(model: llm.LLM) -> str:
    chat_ctx = llm.ChatContext.empty()
    chat_ctx.add_message(role="user", content="Say hello in one sentence.")
    response = await model.chat(chat_ctx=chat_ctx).collect()
    return response.text

Captured Spans

• LiveKit LLMStream requests become Respan chat spans with request messages, completions, tool calls, tool definitions when available, model, provider, and token usage.
• LiveKit execute_function_call executions become Respan tool spans.

Use Respan(..., customer_identifier=..., thread_identifier=..., metadata=...) or respan.propagate_attributes(...) to attach Respan attributes to LiveKit spans.