Module langfuse_interceptor 

Langfuse interceptor for OpenTelemetry-based LLM observability.

This interceptor automatically instruments OpenAI API calls with OpenTelemetry spans. You must configure the OpenTelemetry tracer with Langfuse exporter separately.

Usage

// 1. Build Langfuse exporter
let exporter = ExporterBuilder::from_env()?.build()?;

// 2. Create tracer provider
let provider = SdkTracerProvider::builder()
    .with_span_processor(BatchSpanProcessor::builder(exporter, Tokio).build())
    .build();

global::set_tracer_provider(provider.clone());

// 3. Create interceptor with tracer
let tracer = provider.tracer("openai-ergonomic");
let interceptor = LangfuseInterceptor::new(tracer, LangfuseConfig::new());
let client = Client::from_env()?
    .with_interceptor(Box::new(interceptor))
    .build();

// Traces are automatically sent to Langfuse
let response = client.send_chat(client.chat_simple("Hello!")).await?;

Structs

LangfuseConfig

Configuration for the Langfuse interceptor.

LangfuseInterceptor

Langfuse interceptor for OpenTelemetry-based observability.

LangfuseState

State managed by the Langfuse interceptor across the request lifecycle.