LangGraph and LangChain Observability

What is LangChain and LangGraph Observability?

LangChain and LangGraph observability gives you real-time visibility into your AI agent workflows by collecting traces, spans, and logs using OpenTelemetry. This guide shows you how to instrument your Python-based LangChain or LangGraph application and send telemetry to SigNoz, so you can monitor agent reasoning steps, tool invocations, chain executions, and model responses end-to-end.

With full LangChain and LangGraph observability in SigNoz, you can trace every user request from the initial prompt through each reasoning step, tool execution, and final answer. Correlate traces with logs, set alerts for latency or errors, and continuously improve the reliability of your AI applications.

To get started, check out our example LangChain trip planner agent with OpenTelemetry-based observability/monitoring (via OpenInference). View the LangChain trip planner agent repository.

You can also check out our LangChain SigNoz MCP agent repository.

Prerequisites

A Python application using Python 3.8+
LangChain/LangGraph integrated into your app
Basic understanding of AI Agents and tool calling workflow
SigNoz setup (choose one):
- SigNoz Cloud account with an active ingestion key
- Self-Hosted SigNoz instance
pip installed for managing Python packages
Internet access to send telemetry data to SigNoz Cloud
(Optional but recommended) A Python virtual environment to isolate dependencies

Instrument LangChain and LangGraph with OpenTelemetry

To capture detailed telemetry from LangChain/LangGraph without modifying your core application logic, we use OpenInference, a community-driven standard that provides pre-built instrumentation for popular AI frameworks like LangChain, built on top of OpenTelemetry. This allows you to trace your LangChain application with minimal configuration.

Check out the openinference-instrumentation-langchain package on PyPI for detailed setup instructions.

No-code auto-instrumentation is recommended for quick setup with minimal code changes. It's ideal when you want to get observability up and running without modifying your application code and are leveraging standard instrumentor libraries.

Step 1: Install the necessary packages in your Python environment.

pip install \
  opentelemetry-distro \
  opentelemetry-exporter-otlp \
  opentelemetry-instrumentation-httpx \
  opentelemetry-instrumentation-system-metrics \
  langgraph \
  langchain \
  openinference-instrumentation-langchain

Step 2: Add Automatic Instrumentation

opentelemetry-bootstrap --action=install

Step 3: Configure logging level

To ensure logs are properly captured and exported, configure the root logger to emit logs at the INFO level or higher:

import logging
 
logging.getLogger().setLevel(logging.INFO)

This sets the minimum log level for the root logger to INFO, which ensures that logger.info() calls and higher severity logs (WARNING, ERROR, CRITICAL) are captured by the OpenTelemetry logging auto-instrumentation and sent to SigNoz.

Step 4: Run an example

from langchain.agents import create_agent
 
def add_numbers(a: int, b: int) -> int:
    """Add two numbers together and return the result."""
    return a + b
 
agent = create_agent(
    model="openai:gpt-5-mini",
    tools=[add_numbers],
    system_prompt="You are a helpful math tutor who can do calculations using the provided tools.",
)
 
# Run the agent
agent.invoke(
    {"messages": [{"role": "user", "content": "what is 42 + 58?"}]},
)

📌 Note: Ensure that the OPENAI_API_KEY environment variable is properly defined with your API key before running the code.

Step 5: Run your application with auto-instrumentation

OTEL_RESOURCE_ATTRIBUTES="service.name=<service_name>" \
OTEL_EXPORTER_OTLP_ENDPOINT="https://ingest.<region>.signoz.cloud:443" \
OTEL_EXPORTER_OTLP_HEADERS="signoz-ingestion-key=<your-ingestion-key>" \
OTEL_EXPORTER_OTLP_PROTOCOL=grpc \
OTEL_TRACES_EXPORTER=otlp \
OTEL_METRICS_EXPORTER=otlp \
OTEL_LOGS_EXPORTER=otlp \
OTEL_PYTHON_LOG_CORRELATION=true \
OTEL_PYTHON_LOGGING_AUTO_INSTRUMENTATION_ENABLED=true \
opentelemetry-instrument <your_run_command>

<service_name> is the name of your service
<region>: Your SigNoz Cloud region
<your-ingestion-key>: Your SigNoz ingestion key
Replace <your_run_command> with the actual command you would use to run your application. For example: python main.py

Using Self-Hosted SigNoz? Most steps are identical. To adapt this guide, update the endpoint and remove the ingestion key header as shown in Cloud → Self-Hosted.

Code-based instrumentation gives you fine-grained control over your telemetry configuration. Use this approach when you need to customize resource attributes, sampling strategies, or integrate with existing observability infrastructure.

Step 1: Install OpenInference and OpenTelemetry related packages

pip install openinference-instrumentation-langchain \
opentelemetry-exporter-otlp \
opentelemetry-sdk \
langgraph \
langchain

Step 2: Import the necessary modules in your Python application

from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.resources import Resource
from openinference.instrumentation.langchain import LangChainInstrumentor

Step 3: Set up the OpenTelemetry Tracer Provider to send traces directly to SigNoz Cloud

resource = Resource.create({"service.name": "<service_name>"})
provider = TracerProvider(resource=resource)
span_exporter = OTLPSpanExporter(
    endpoint="https://ingest.<region>.signoz.cloud:443/v1/traces",
    headers={"signoz-ingestion-key": "<your-ingestion-key>"},
)
provider.add_span_processor(BatchSpanProcessor(span_exporter))

<service_name> is the name of your service
Set the <region> to match your SigNoz Cloud region
Replace <your-ingestion-key> with your SigNoz ingestion key

Using Self-Hosted SigNoz? Most steps are identical. To adapt this guide, update the endpoint and remove the ingestion key header as shown in Cloud → Self-Hosted.

Step 4: Instrument LangChain using OpenInference

Use the LangChainInstrumentor from OpenInference to automatically trace LangChain operations with your OpenTelemetry setup:

LangChainInstrumentor().instrument()

📌 Important: Place this code at the start of your application logic — before any LangChain/LangGraph functions are called or used — to ensure telemetry is correctly captured.

Step 5: Run an example

from langchain.agents import create_agent
 
def add_numbers(a: int, b: int) -> int:
    """Add two numbers together and return the result."""
    return a + b
 
agent = create_agent(
    model="openai:gpt-5-mini",
    tools=[add_numbers],
    system_prompt="You are a helpful math tutor who can do calculations using the provided tools.",
)
 
# Run the agent
agent.invoke(
    {"messages": [{"role": "user", "content": "what is 42 + 58?"}]},
)

📌 Note: Ensure that the OPENAI_API_KEY environment variable is properly defined with your API key before running the code.

Once configured, LangChain and LangGraph automatically emit traces, spans, and attributes for every agent request.

LangChain and LangGraph traces are available in SigNoz under the Traces tab:

Traces View — *Traces of your LangChain Application*

When you click on a trace ID in SigNoz, you'll see a detailed view of the trace, including all associated spans, along with their events and attributes:

Detailed Traces View — *Detailed traces view of your LangChain Application*

Use the SigNoz MCP to automatically recognize important LangChain and LangGraph trace attributes and create a custom dashboard using natural language, with no manual configuration needed.

Send Traces with the LangSmith OpenTelemetry Exporter (Optional)

If your app already uses LangSmith, the LangSmith SDK can send LangChain and LangGraph traces to SigNoz over OTLP. You do not need a LangSmith API key. Do not combine this path with the OpenInference steps above. If LangChainInstrumentor().instrument() also runs, only the OpenInference spans reach SigNoz, so remove it first.

Step 1: Install the packages

pip install "langsmith[otel]>=0.4.25" langchain langchain-openai langgraph

Step 2: Create an example app

The example calls OpenAI, so set the OPENAI_API_KEY environment variable before you run it.

from typing import TypedDict
 
from langchain_openai import ChatOpenAI
from langgraph.graph import END, START, StateGraph
 
llm = ChatOpenAI(model="gpt-4.1-mini")
 
 
class State(TypedDict):
    question: str
    answer: str
 
 
def call_model(state: State) -> dict:
    return {"answer": llm.invoke(state["question"]).content}
 
 
graph = StateGraph(State)
graph.add_node("call_model", call_model)
graph.add_edge(START, "call_model")
graph.add_edge("call_model", END)
app = graph.compile()
 
print(app.invoke({"question": "Name one city in France."})["answer"])

Step 3: Run your application with the LangSmith OpenTelemetry exporter

LANGSMITH_TRACING=true \
LANGSMITH_OTEL_ENABLED=true \
LANGSMITH_OTEL_ONLY=true \
OTEL_SERVICE_NAME=<service_name> \
OTEL_EXPORTER_OTLP_ENDPOINT="https://ingest.<region>.signoz.cloud:443" \
OTEL_EXPORTER_OTLP_HEADERS="signoz-ingestion-key=<your-ingestion-key>" \
python main.py

Verify these values:

<service_name>: the name of your service
<region>: your SigNoz Cloud region
<your-ingestion-key>: your SigNoz ingestion key

LANGSMITH_OTEL_ONLY=true sends traces to SigNoz only. Without it, the SDK also sends traces to LangSmith, which needs a LangSmith API key. Enter the endpoint without a path, because the exporter adds /v1/traces. For Self-Hosted SigNoz, set OTEL_EXPORTER_OTLP_ENDPOINT to http://<signoz-host>:4318 and remove OTEL_EXPORTER_OTLP_HEADERS. The LangSmith SDK exports over OTLP/HTTP, so use port 4318, not 4317.

Each graph run creates a LangGraph span, a span for each node, and a ChatOpenAI span for the model call. In AI Observability, the model call shows its token counts, estimated cost, and messages. In Self-Hosted SigNoz, the cost shows only after you add a price for the model. With a custom collector configuration, also add the AI observability processors.

A LangGraph trace in SigNoz with the LangGraph, call_model, and ChatOpenAI spans, and the ChatOpenAI span details with tokens, cost, input, and output — *A LangGraph trace from the LangSmith OpenTelemetry exporter*

The Thread tab of a LangGraph trace in SigNoz, with the user message, the assistant reply, and the finish reason — *The Thread tab shows the conversation of the trace*

LangChain and LangGraph Observability in JavaScript

You can instrument your LangChain/LangGraph applications in JavaScript using the OpenInference LangChain Instrumentor package.

For detailed guidance on instrumenting JavaScript applications with OpenTelemetry and connecting them to SigNoz, see the SigNoz OpenTelemetry JavaScript instrumentation docs.

Trace the rest of your LLM application and retrieval stack:

LlamaIndex observability with OpenTelemetry - trace RAG queries, retrievers, and index operations
Monitor Haystack pipelines with OpenTelemetry - trace RAG pipelines, retrievers, and generator nodes
DSPy observability with OpenTelemetry - trace modules, chains, and optimizer runs
Google ADK observability with OpenTelemetry - trace ADK agents, tools, and the model calls they make
Monitor Firecrawl with OpenTelemetry - trace scrape, crawl, search, and map calls alongside credit usage

LangChain & LangGraph Observability with OpenTelemetry

What is LangChain and LangGraph Observability?

Prerequisites

Instrument LangChain and LangGraph with OpenTelemetry

Send Traces with the LangSmith OpenTelemetry Exporter (Optional)

LangChain and LangGraph Observability in JavaScript

Related integrations