import base64
import os
from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.tools.duckduckgo import DuckDuckGoTools
LANGFUSE_AUTH = base64.b64encode(
f"{os.getenv('LANGFUSE_PUBLIC_KEY')}:{os.getenv('LANGFUSE_SECRET_KEY')}".encode()
).decode()
os.environ["OTEL_EXPORTER_OTLP_ENDPOINT"] = (
"https://us.cloud.langfuse.com/api/public/otel" # πΊπΈ US data region
)
# os.environ["OTEL_EXPORTER_OTLP_ENDPOINT"]="https://cloud.langfuse.com/api/public/otel" # πͺπΊ EU data region
# os.environ["OTEL_EXPORTER_OTLP_ENDPOINT"]="http://localhost:3000/api/public/otel" # π Local deployment (>= v3.22.0)
os.environ["OTEL_EXPORTER_OTLP_HEADERS"] = f"Authorization=Basic {LANGFUSE_AUTH}"
from opentelemetry.exporter.otlp.proto.http.trace_exporter import ( # noqa: E402
OTLPSpanExporter,
)
from opentelemetry.sdk.trace import TracerProvider # noqa: E402
from opentelemetry.sdk.trace.export import SimpleSpanProcessor # noqa: E402
trace_provider = TracerProvider()
trace_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter()))
# Sets the global default tracer provider
from opentelemetry import trace # noqa: E402
trace.set_tracer_provider(trace_provider)
# Creates a tracer from the global tracer provider
tracer = trace.get_tracer(__name__)
import openlit # noqa: E402
# Initialize OpenLIT instrumentation. The disable_batch flag is set to true to process traces immediately.
openlit.init(tracer=tracer, disable_batch=True)
agent = Agent(
model=OpenAIChat(id="gpt-5-mini"),
tools=[DuckDuckGoTools()],
markdown=True,
debug_mode=True,
)
agent.print_response("What is currently trending on Twitter?")
Create a virtual environment
Terminal
and create a python virtual environment.python3 -m venv .venv
source .venv/bin/activate
Set your API key
export LANGFUSE_PUBLIC_KEY=<your-key>
export LANGFUSE_SECRET_KEY=<your-key>
Install libraries
pip install -U agno openai langfuse openlit opentelemetry-sdk opentelemetry-exporter-otlp ddgs
Run Agent
python cookbook/integrations/observability/langfuse_via_openlit.py