> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fact0.io/llms.txt
> Use this file to discover all available pages before exploring further.

# LangGraph

> Trace stateful, multi-step agent graphs with node-level telemetry spans.

# LangGraph

<Note>
  Native LangGraph helpers are on the roadmap. Use the manual pattern below today - it works with any `StateGraph`.
</Note>

## Pattern

Wrap each node with a telemetry span and reuse the same `execution_id` across nodes.

```python theme={null}
import fact0
from langgraph.graph import StateGraph

client = fact0.Client(api_key="f0_live_...")

def with_telemetry(name: str, fn):
    def wrapped(state):
        with client.telemetry.execution(
            agent_id=state.get("agent_id", "graph"),
            execution_id=state["execution_id"],
        ) as ex:
            with ex.span(name, span_type="CUSTOM") as span:
                result = fn(state)
                span.complete(output=result)
                return result
    return wrapped

graph = StateGraph(dict)
graph.add_node("plan", with_telemetry("plan", plan_fn))
graph.add_node("act", with_telemetry("act", act_fn))
graph.add_node("reflect", with_telemetry("reflect", reflect_fn))
```

## What gets captured

* One execution per graph run (shared `execution_id`)
* One `CUSTOM` span per node, with input/output state
* Use `MODEL_INVOCATION` / `TOOL_CALL` span types inside nodes that call LLMs or external tools

## Related

* [LangChain](/integrations/langchain) - callback-based instrumentation
* [Telemetry client](/sdk/python/telemetry)