LangGraph

Import name: Sentry.instrumentLangGraph

The instrumentLangGraph helper adds instrumentation for @langchain/langgraph by wrapping a StateGraph before compilation and recording AI agent interactions with configurable input/output capture. The OpenTelemetry-based automatic integration available for Node.js does not work in React Native, so calling instrumentLangGraph on the graph before .compile() is the only supported path.

Usage

import * as Sentry from "@sentry/react-native";
import { ChatOpenAI } from "@langchain/openai";
import {
  StateGraph,
  MessagesAnnotation,
  START,
  END,
} from "@langchain/langgraph";
import { SystemMessage, HumanMessage } from "@langchain/core/messages";

const llm = new ChatOpenAI({
  modelName: "gpt-4o",
  // Warning: API keys included in your app bundle will be visible to anyone who
  // inspects the bundle. Proxy LLM calls through your own backend whenever possible.
  apiKey: "your-api-key",
});

async function callLLM(state) {
  const response = await llm.invoke(state.messages);
  return {
    messages: [...state.messages, response],
  };
}

const agent = new StateGraph(MessagesAnnotation)
  .addNode("agent", callLLM)
  .addEdge(START, "agent")
  .addEdge("agent", END);

// Instrument the graph BEFORE compiling
Sentry.instrumentLangGraph(agent, {
  recordInputs: true,
  recordOutputs: true,
});

const graph = agent.compile({ name: "my_agent" });

const result = await graph.invoke({
  messages: [
    new SystemMessage("You are a helpful assistant."),
    new HumanMessage("Hello!"),
  ],
});

import * as Sentry from "@sentry/react-native";
import { ChatOpenAI } from "@langchain/openai";
import {
  StateGraph,
  MessagesAnnotation,
  START,
  END,
} from "@langchain/langgraph";
import { SystemMessage, HumanMessage } from "@langchain/core/messages";

const llm = new ChatOpenAI({
  modelName: "gpt-4o",
  // Warning: API keys included in your app bundle will be visible to anyone who
  // inspects the bundle. Proxy LLM calls through your own backend whenever possible.
  apiKey: "your-api-key",
});

async function callLLM(state) {
  const response = await llm.invoke(state.messages);
  return {
    messages: [...state.messages, response],
  };
}

const agent = new StateGraph(MessagesAnnotation)
  .addNode("agent", callLLM)
  .addEdge(START, "agent")
  .addEdge("agent", END);

// Instrument the graph BEFORE compiling
Sentry.instrumentLangGraph(agent, {
  recordInputs: true,
  recordOutputs: true,
});

const graph = agent.compile({ name: "my_agent" });

const result = await graph.invoke({
  messages: [
    new SystemMessage("You are a helpful assistant."),
    new HumanMessage("Hello!"),
  ],
});

Make sure tracing is enabled for the spans produced by this integration to be captured.

Configuration

Options

The following options control what data is captured from LangGraph operations:

recordInputs

Type: boolean (optional)

Records inputs to LangGraph operations (such as prompts and messages).

Defaults to true if sendDefaultPii is true.

recordOutputs

Type: boolean (optional)

Records outputs from LangGraph operations (such as generated text and responses).

Defaults to true if sendDefaultPii is true.

Supported Operations

By default, tracing support is added to the following LangGraph SDK calls:

• Agent Creation (gen_ai.create_agent) - Captures spans when compiling a StateGraph into an executable agent
• Agent Invocation (gen_ai.invoke_agent) - Captures spans for agent execution via invoke()

Supported Versions

• @langchain/langgraph: >=0.2.0 <2.0.0