> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agno.com/llms.txt
> Use this file to discover all available pages before exploring further.

# LangGraph

> Wrap a compiled LangGraph graph as an AgentOS agent.

`LangGraphAgent` wraps a compiled LangGraph graph so it can be served through AgentOS or used standalone.

```python theme={null}
from agno.agents.langgraph import LangGraphAgent
from agno.db.sqlite import SqliteDb
from agno.os import AgentOS
from langchain_openai import ChatOpenAI
from langgraph.graph import MessagesState, StateGraph


def chatbot(state: MessagesState):
    return {"messages": [ChatOpenAI(model="gpt-5.4").invoke(state["messages"])]}


graph = StateGraph(MessagesState)
graph.add_node("chatbot", chatbot)
graph.set_entry_point("chatbot")
compiled = graph.compile()

agent = LangGraphAgent(
    name="LangGraph Chatbot",
    graph=compiled,
)

agent_os = AgentOS(
    agents=[agent],
    tracing=True,
    db=SqliteDb(db_file="tmp/agentos.db"),
)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="langgraph_agent:app", reload=True)
```

## Install

```bash theme={null}
uv pip install langgraph langchain-openai
export OPENAI_API_KEY=sk-...
```

## Parameters

| Parameter    | Type             | Default      | Description                                                   |
| ------------ | ---------------- | ------------ | ------------------------------------------------------------- |
| `name`       | `str`            | `None`       | Display name for the agent.                                   |
| `id`         | `str`            | `None`       | Unique identifier. Auto-generated from `name` if unset.       |
| `graph`      | `CompiledGraph`  | `None`       | Compiled LangGraph graph (`graph.compile()`).                 |
| `input_key`  | `str`            | `"messages"` | Key in the graph state used for input messages.               |
| `output_key` | `str`            | `"messages"` | Key in the graph state used for output messages.              |
| `config`     | `Dict[str, Any]` | `None`       | Optional LangGraph config dict passed to `invoke` / `stream`. |
| `db`         | `BaseDb`         | `None`       | Database for session persistence.                             |

## Examples

<CardGroup cols={2}>
  <Card title="AgentOS deployment" icon="server" href="https://github.com/agno-agi/agno/blob/main/cookbook/frameworks/00_quickstart/langgraph_agent.py">
    Serve a compiled LangGraph through AgentOS.
  </Card>

  <Card title="Standalone usage" icon="play" href="https://github.com/agno-agi/agno/blob/main/cookbook/frameworks/langgraph/langgraph_basic.py">
    Call the agent directly with `.run()` and `.print_response()`.
  </Card>

  <Card title="Graphs with tools" icon="wrench" href="https://github.com/agno-agi/agno/blob/main/cookbook/frameworks/langgraph/langgraph_tools.py">
    Tool nodes that surface as Agno tool events.
  </Card>

  <Card title="Sessions" icon="comments" href="https://github.com/agno-agi/agno/blob/main/cookbook/frameworks/langgraph/langgraph_session.py">
    Resume conversations across runs with `session_id`.
  </Card>

  <Card title="Time travel" icon="clock-rotate-left" href="https://github.com/agno-agi/agno/blob/main/cookbook/frameworks/langgraph/langgraph_time_travel.py">
    Replay and fork runs from LangGraph checkpoints.
  </Card>
</CardGroup>

## Developer Resources

* [Cookbook examples](https://github.com/agno-agi/agno/tree/main/cookbook/frameworks/langgraph)
* [LangGraph documentation](https://langchain-ai.github.io/langgraph/)