LangGraph

LangGraph is a library for building stateful, multi-actor applications with LLMs. SideSeat captures runs from graph steps so you can debug state transitions.

Prerequisites

• SideSeat running locally (sideseat)
• Python 3.9+
• OpenAI API key (or another LangChain-supported model)

Quick Start

1. Start SideSeat

   npx sideseat

2. Install dependencies

   pip

   pip install langgraph langchain-openai "sideseat[langgraph]"

   uv

   uv add langgraph langchain-openai "sideseat[langgraph]"

3. Add telemetry

   from sideseat import SideSeat, Frameworks
   from langgraph.prebuilt import create_react_agent
   from langchain_openai import ChatOpenAI
   
   SideSeat(framework=Frameworks.LangGraph)
   
   llm = ChatOpenAI(model="gpt-5-mini")
   agent = create_react_agent(llm, tools=[])
   result = agent.invoke({"messages": [("user", "Hello!")]})
   print(result["messages"][-1].content)

4. View runs

   Open http://localhost:5388 to see your runs.

Without SideSeat SDK

1. Start SideSeat

   npx sideseat

2. Set the endpoint

   export OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:5388/otel/default

3. Install dependencies

   pip

   pip install langgraph langchain-openai openinference-instrumentation-langchain opentelemetry-exporter-otlp

   uv

   uv add langgraph langchain-openai openinference-instrumentation-langchain opentelemetry-exporter-otlp

4. Add telemetry

   from opentelemetry import trace
   from opentelemetry.sdk.trace import TracerProvider
   from opentelemetry.sdk.trace.export import BatchSpanProcessor
   from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
   from openinference.instrumentation.langchain import LangChainInstrumentor
   
   provider = TracerProvider()
   provider.add_span_processor(BatchSpanProcessor(OTLPSpanExporter()))
   trace.set_tracer_provider(provider)
   LangChainInstrumentor().instrument(tracer_provider=provider, skip_dep_check=True)

5. View runs

   Open http://localhost:5388 to see your runs.

What You’ll See

SideSeat shows:

• Parent span for the graph execution
• Child spans for each node invocation
• LLM call spans with token counts and costs

Tools

Add tools to your agent:

from langchain_core.tools import tool

@tool
def get_weather(location: str) -> str:
    """Get weather for a location."""
    return f"Sunny in {location}"

llm = ChatOpenAI(model="gpt-5-mini")
agent = create_react_agent(llm, tools=[get_weather])
result = agent.invoke({"messages": [("user", "What's the weather in Paris?")]})

Complete Example

from sideseat import SideSeat, Frameworks
from typing import TypedDict, Annotated
from langgraph.graph import StateGraph, END
from langgraph.graph.message import add_messages
from langchain_openai import ChatOpenAI

SideSeat(framework=Frameworks.LangGraph)

class State(TypedDict):
    messages: Annotated[list, add_messages]

llm = ChatOpenAI(model="gpt-5-mini")

def chatbot(state: State):
    return {"messages": [llm.invoke(state["messages"])]}

graph = StateGraph(State)
graph.add_node("chatbot", chatbot)
graph.set_entry_point("chatbot")
graph.add_edge("chatbot", END)

app = graph.compile()
result = app.invoke({"messages": [("user", "What is 2+2?")]})
print(result["messages"][-1].content)

Next Steps

• Python SDK — SDK reference
• Core Concepts — understanding runs, steps, and messages