LangGraph - Graph-based orchestration framework for building stateful LLM applications

LangGraph provides a low-level graph orchestration framework that enables:

• Stateful, long-running agent workflows
• Conditional routing and branching
• Subgraph composition
• Cycle support for iterative processes

Core Concepts

• StateGraph: Main graph class that manages State, Nodes, and Edges
• Node: Execution unit that takes state and produces state updates
• Edge: Transition between nodes (fixed or conditional)
• State: Data structure that flows through the graph

Example

use lc_langgraph::{StateGraph, GraphNode, StateSchema, START, END};
use serde::{Deserialize, Serialize};

// Define state
#[derive(Serialize, Deserialize, Clone)]
struct MyState {
    messages: Vec<String>,
    count: usize,
}

impl StateSchema for MyState {}

// Create graph
let mut graph = StateGraph::<MyState>::new();

// Add nodes
graph.add_node("process", |state: MyState| {
    MyState {
        messages: state.messages.clone(),
        count: state.count + 1,
    }
});

// Add edges
graph.add_edge(START, "process");
graph.add_edge("process", END);

// Compile and run
let compiled = graph.compile();
let result = compiled.invoke(MyState { messages: vec![], count: 0 }).await?;