Skip to main content
Prefer zero setup? Use plog run or import provenlog.auto instead. See Auto-Instrumentation.

​
Setup

from provenlog.integrations.langgraph import Trail

trail = Trail(agent_id="my-graph-agent")
graph.invoke(input, config={"callbacks": [trail]})

​
What gets captured

Everything from the LangChain integration, plus graph-specific events:
EventAction TypeDetails
Node startCUSTOMNode name, step index
Node endCUSTOMNode name, output
Node errorCUSTOMNode name, error details
Graph stateCUSTOMCheckpoint metadata

​
How it works

The LangGraph Trail extends the LangChain callback handler with graph-node-level awareness. It tracks which node is executing at any point, so LLM calls and tool calls within a graph are attributed to their specific node.
When auto-instrumentation detects LangGraph, it skips the separate LangChain patch — the LangGraph Trail captures both graph-level and chain-level events.

​
Configuration

# Simple — uses default embedded mode
trail = Trail(agent_id="my-graph-agent")

# With explicit client for custom configuration
from provenlog import ProvenLogClient

client = ProvenLogClient("http://localhost:7600", agent_id="my-agent")
trail = Trail(client=client, agent_id="my-graph-agent")