# OpenTelemetry Tracing - Python Code for the steps in `shared/opentelemetry.md`. Read that file first: it covers what to check before adding tracing, the install table, the option table, content modes, API tracing and its destinations, and the span reference. ## Register a tracer provider This example prints each span to the console. Register the tracer provider before the first call that should be traced. ```python from anthropic import Anthropic from opentelemetry import trace from opentelemetry.sdk.trace import TracerProvider from opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor provider = TracerProvider() provider.add_span_processor(SimpleSpanProcessor(ConsoleSpanExporter())) trace.set_tracer_provider(provider) client = Anthropic() message = client.messages.create( model="{{OPUS_ID}}", max_tokens=1024, messages=[{"role": "user", "content": "Hello, Claude"}], ) ``` ## Configure the SDK ```python # Use a specific tracer provider and record content. client = Anthropic( open_telemetry={"tracer_provider": provider, "traces": {"content_mode": "content"}} ) # Turn spans off for one client. quiet_client = Anthropic(open_telemetry=False) # Conversation ID for one call. with_options() returns a copy of the client and # REPLACES the whole option, so start from client.open_telemetry. message = client.with_options( open_telemetry={**client.open_telemetry, "conversation_id": "conv_42"} ).messages.create( model="{{OPUS_ID}}", max_tokens=1024, messages=[{"role": "user", "content": "Hello, Claude"}], ) ``` ## Connect API spans to your own traces Call the SDK from inside one of your own spans, so your span contains the SDK span, which contains the API span. ```python from opentelemetry import trace tracer = trace.get_tracer("my-service") with tracer.start_as_current_span("summarize_issues"): client.beta.sessions.events.send( session.id, events=[{"type": "user.message", "content": [{"type": "text", "text": "Summarize the open issues"}]}], ) ``` Without the SDK (raw HTTP or another client), inject the active span's trace context into the request headers: ```python import os from opentelemetry.trace.propagation.tracecontext import TraceContextTextMapPropagator headers = { "x-api-key": os.environ["ANTHROPIC_API_KEY"], "anthropic-version": "2023-06-01", "content-type": "application/json", } TraceContextTextMapPropagator().inject(headers) # adds traceparent for the active span ``` After the SDK setup, offer setting up Claude API tracing where `shared/opentelemetry.md` -> SDK tracing vs API tracing says to, if you didn't already.