Skip to main content

LangGraph Integration

Coming Soon — LangGraph integration is currently under development.

Overview

The MutagenT LangGraph integration provides tracing for graph-based agent workflows, capturing node executions, state transitions, and tool calls.

Expected Installation

bun add @mutagent/langgraph

Expected Usage

import { MutagentTracer } from '@mutagent/langgraph';
import { StateGraph } from '@langchain/langgraph';

// Create the tracer
const tracer = new MutagentTracer({
  apiKey: process.env.MUTAGENT_API_KEY,
});

// Create your graph
const graph = new StateGraph({...})
  .addNode('agent', agentNode)
  .addNode('tools', toolNode)
  .addEdge('agent', 'tools');

// Compile with tracer
const app = graph.compile({
  callbacks: [tracer],
});

// Run with automatic tracing
const result = await app.invoke({
  messages: [{ role: 'user', content: 'Hello' }],
});

What Gets Tracked

  • Graph Execution: Full workflow visualization
  • Node Execution: Each node’s inputs/outputs
  • State Transitions: State changes between nodes
  • Conditional Edges: Decision points and routing
  • Tool Calls: External tool invocations
  • Subgraphs: Nested graph executions

Visualization

Traces appear in MutagenT with:
  • Graph structure visualization
  • Node-by-node execution timeline
  • State diff at each transition
  • Error highlighting
[AI AGENT PLACEHOLDER: LANGGRAPH INTEGRATION WILL BE DOCUMENTED WHEN AVAILABLE]