Integrate Langfuse with Semantic Kernel
This notebook demonstrates how to integrate Langfuse with Semantic Kernel for improved observability and debugging. By the end of this notebook, you will be able to trace your Semantic Kernel applications with Langfuse.
What is Semantic Kernel? Semantic Kernel (GitHub) is an open-source SDK developed by Microsoft that combines LLMs with programming languages like C#, Python, and Java. It allows developers to create AI applications by integrating AI services, data sources, and logic, enabling rapid delivery of enterprise-grade solutions.
What is Langfuse? Langfuse is an open-source platform for LLM observability. It provides tracing and monitoring capabilities for AI applications, helping developers debug, analyze, and optimize their AI systems. Langfuse integrates with various tools and frameworks via native integrations, OpenTelemetry, and SDKs.
Get Started
We’ll walk through a simple example of using Semantic Kernel and integrating it with Langfuse.
Step 1: Install Dependencies
Install the necessary packages:
%pip install langfuse openlit semantic-kernelStep 2: Set Up Environment Variables
Set your Langfuse API keys and configure the Langfuse SDK. Replace the placeholders with your actual API keys.
import os
import base64
# Get your own keys from https://cloud.langfuse.com
os.environ["LANGFUSE_PUBLIC_KEY"] = "pk-lf-..."
os.environ["LANGFUSE_SECRET_KEY"] = "sk-lf-..."
os.environ["LANGFUSE_HOST"] = "https://cloud.langfuse.com" # 🇪🇺 EU region example
# os.environ["LANGFUSE_HOST"] = "https://us.cloud.langfuse.com" # 🇺🇸 US region example
LANGFUSE_AUTH = base64.b64encode(
f"{os.environ.get('LANGFUSE_PUBLIC_KEY')}:{os.environ.get('LANGFUSE_SECRET_KEY')}".encode()
).decode()
os.environ["OTEL_EXPORTER_OTLP_ENDPOINT"] = os.environ.get("LANGFUSE_HOST") + "/api/public/otel"
os.environ["OTEL_EXPORTER_OTLP_HEADERS"] = f"Authorization=Basic {LANGFUSE_AUTH}"
# your openai key
os.environ["OPENAI_API_KEY"] = "sk-proj-..."
os.environ["OPENAI_CHAT_MODEL_ID"] = "gpt-4o"Configure tracer_provider and add a span processor to export traces to Langfuse. OTLPSpanExporter() uses the endpoint and headers from the environment variables.
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace.export import SimpleSpanProcessor
trace_provider = TracerProvider()
trace_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter()))
# Sets the global default tracer provider
from opentelemetry import trace
trace.set_tracer_provider(trace_provider)
# Creates a tracer from the global tracer provider
tracer = trace.get_tracer(__name__)Step 3: Initialize OpenLit
Initialize the OpenLit instrumentation SDK to start capturing OpenTelemetry traces.
import openlit
# Initialize OpenLIT instrumentation. The disable_batch flag is set to true to process traces immediately.
openlit.init(tracer=tracer, disable_batch=True)Step 4: Create a Simple Semantic Kernel Application
We’ll create a simple Semantic Kernel application where an Assistant agent answers a user’s question.
from semantic_kernel import Kernel
kernel = Kernel()from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
kernel.add_service(
OpenAIChatCompletion(),
)from semantic_kernel.connectors.ai.open_ai import AzureChatPromptExecutionSettings, OpenAIChatPromptExecutionSettings
from semantic_kernel.prompt_template import InputVariable, PromptTemplateConfig
prompt = """{{$input}}
Answer the question above.
"""
prompt_template_config = PromptTemplateConfig(
template=prompt,
name="summarize",
template_format="semantic-kernel",
input_variables=[
InputVariable(name="input", description="The user input", is_required=True),
]
)
summarize = kernel.add_function(
function_name="summarizeFunc",
plugin_name="summarizePlugin",
prompt_template_config=prompt_template_config,
)input_text = "What is Langfuse?"
summary = await kernel.invoke(summarize, input=input_text)
print(summary)Step 5: Pass Additional Attributes (Optional)
Opentelemetry lets you attach a set of attributes to all spans by setting set_attribute. This allows you to set properties like a Langfuse Session ID, to group traces into Langfuse Sessions or a User ID, to assign traces to a specific user. You can find a list of all supported attributes in the here.
with tracer.start_as_current_span("Semantic-Kernel-Trace") as span:
span.set_attribute("langfuse.user.id", "user-123")
span.set_attribute("langfuse.session.id", "123456789")
span.set_attribute("langfuse.tags", ["semantic-kernel", "demo"])
span.set_attribute("langfuse.prompt.name", "test-1")
input = "What is Langfuse?"
output = await kernel.invoke(summarize, input=input_text)
print(summary)Alternatively, OpenTelemetry traces in Langfuse can also be modified using the Python low-level SDK. For this, we create a new parent span and fetch the OpenTelemetry trace_id. This trace_id is then used to modify the span. Have a look at the Python low-level SDK for more examples.
from opentelemetry.trace import format_trace_id
from langfuse import Langfuse
langfuse = Langfuse()
with tracer.start_as_current_span("Semantic-Kernel-Trace") as span:
input = "What is Langfuse?"
output = await kernel.invoke(summarize, input=input_text)
print(output)
# Get the trace_id from the Otel span
current_span = trace.get_current_span()
span_context = current_span.get_span_context()
trace_id = span_context.trace_id
formatted_trace_id = format_trace_id(trace_id)
# Update the trace using the low-level Python SDK.
langfuse.trace(
id=formatted_trace_id,
input=input,
output=output,
name = "docs-retrieval",
user_id = "user__935d7d1d-8625-4ef4-8651-544613e7bd22",
metadata = {"email": "[email protected]"},
tags = ["production"]
)Step 6: See Traces in Langfuse
After running the agent above, you can log in to your Langfuse dashboard and view the traces generated by your Semantic Kernel application. Here is an example screenshot of a trace in Langfuse:
You can also view the public trace here: Langfuse Trace Example