LangGraph

LangGraph extends LangChain with a graph-based runtime for building stateful agent and multi-actor workflows. Unlike linear chains, LangGraph supports cycles, conditional branching, parallel execution, and durable persistence — making it the go-to framework for production agentic systems.

Key features

• Stateful graphs — each node reads and writes to a shared typed state object
• Cycles — agents can loop, retry, and re-plan — not just run top-to-bottom
• Human-in-the-loop — pause execution at any node and resume after human review
• Persistence — checkpoint state to SQLite, Postgres, or Redis for fault tolerance
• Multi-agent — route messages between specialized sub-graphs (supervisor, swarm patterns)
• LangGraph Platform — hosted deployment with streaming, webhooks, and a Studio debugger

Quick start

npx ai-supply add langgraph-stateful-agent-workflows

# Or install directly
pip install langgraph langchain-anthropic

from typing import TypedDict
from langgraph.graph import StateGraph, END
from langchain_anthropic import ChatAnthropic

class State(TypedDict):
    messages: list

llm = ChatAnthropic(model="claude-opus-4-5")

def call_model(state: State):
    response = llm.invoke(state["messages"])
    return {"messages": state["messages"] + [response]}

graph = StateGraph(State)
graph.add_node("agent", call_model)
graph.set_entry_point("agent")
graph.add_edge("agent", END)

app = graph.compile()
result = app.invoke({"messages": [{"role": "user", "content": "Hello!"}]})
print(result["messages"][-1].content)

Curated mirror of the open-source LangGraph project (MIT). Install upstream from the repository.