🌍 Global Mirror — Visit original CN site →
LangChain Reference home pageLangChain Reference

langgraph

Description

LangGraph Logo LangGraph Logo

Version Downloads Open Issues Docs

Trusted by companies shaping the future of agents – including Klarna, Replit, Elastic, and more – LangGraph is a low-level orchestration framework for building, managing, and deploying long-running, stateful agents.

Get started

Install LangGraph:

pip install -U langgraph
Copy

Create a simple workflow:

from langgraph.graph import START, StateGraph
from typing_extensions import TypedDict

class State(TypedDict):
    text: str

def node_a(state: State) -> dict:
    return {"text": state["text"] + "a"}

def node_b(state: State) -> dict:
    return {"text": state["text"] + "b"}

graph = StateGraph(State)
graph.add_node("node_a", node_a)
graph.add_node("node_b", node_b)
graph.add_edge(START, "node_a")
graph.add_edge("node_a", "node_b")

print(graph.compile().invoke({"text": ""}))
# {'text': 'ab'}
Copy

Get started with the LangGraph Quickstart.

To quickly build agents with LangChain's create_agent (built on LangGraph), see the LangChain Agents documentation.

Core benefits

LangGraph provides low-level supporting infrastructure for any long-running, stateful workflow or agent. LangGraph does not abstract prompts or architecture, and provides the following central benefits:

  • Durable execution: Build agents that persist through failures and can run for extended periods, automatically resuming from exactly where they left off.
  • Human-in-the-loop: Seamlessly incorporate human oversight by inspecting and modifying agent state at any point during execution.
  • Comprehensive memory: Create truly stateful agents with both short-term working memory for ongoing reasoning and long-term persistent memory across sessions.
  • Debugging with LangSmith: Gain deep visibility into complex agent behavior with visualization tools that trace execution paths, capture state transitions, and provide detailed runtime metrics.
  • Production-ready deployment: Deploy sophisticated agent systems confidently with scalable infrastructure designed to handle the unique challenges of stateful, long-running workflows.

LangGraph’s ecosystem

While LangGraph can be used standalone, it also integrates seamlessly with any LangChain product, giving developers a full suite of tools for building agents. To improve your LLM application development, pair LangGraph with:

  • LangSmith — Helpful for agent evals and observability. Debug poor-performing LLM app runs, evaluate agent trajectories, gain visibility in production, and improve performance over time.
  • LangSmith Deployment — Deploy and scale agents effortlessly with a purpose-built deployment platform for long running, stateful workflows. Discover, reuse, configure, and share agents across teams — and iterate quickly with visual prototyping in LangGraph Studio.
  • LangChain – Provides integrations and composable components to streamline LLM application development.

[!NOTE] Looking for the JS version of LangGraph? See the JS repo and the JS docs.

Additional resources

  • Guides: Quick, actionable code snippets for topics such as streaming, adding memory & persistence, and design patterns (e.g. branching, subgraphs, etc.).
  • Reference: Detailed reference on core classes, methods, how to use the graph and checkpointing APIs, and higher-level prebuilt components.
  • Examples: Guided examples on getting started with LangGraph.
  • LangChain Forum: Connect with the community and share all of your technical questions, ideas, and feedback.
  • LangChain Academy: Learn the basics of LangGraph in our free, structured course.
  • Case studies: Hear how industry leaders use LangGraph to ship AI applications at scale.

Acknowledgements

LangGraph is inspired by Pregel and Apache Beam. The public interface draws inspiration from NetworkX. LangGraph is built by LangChain Inc, the creators of LangChain, but can be used without LangChain.

Classes

Class

ErrorCode

Class

GraphRecursionError

Raised when the graph has exhausted the maximum number of steps.

Class

InvalidUpdateError

Raised when attempting to update a channel with an invalid set of updates.

Class

GraphBubbleUp

Class

GraphInterrupt

Raised when a subgraph is interrupted, suppressed by the root graph.

Class

ParentCommand

Class

EmptyInputError

Raised when graph receives an empty input.

Class

TaskNotFound

Raised when the executor is unable to find a task (for distributed mode).

Class

Runtime

Convenience class that bundles run-scoped context and other runtime utilities.

Class

ToolOutputMixin

Class

RetryPolicy

Configuration for retrying nodes.

Class

CachePolicy

Configuration for caching nodes.

Class

Interrupt

Information about an interrupt that occurred in a node.

Class

StateUpdate

Class

PregelTask

A Pregel task.

Class

CacheKey

Cache key for a task.

Class

PregelExecutableTask

Class

StateSnapshot

Snapshot of the state of the graph at the beginning of a step.

Class

Send

A message or packet to send to a specific node in the graph.

Class

Command

One or more commands to update the graph's state and send messages to nodes.

Class

Overwrite

Bypass a reducer and write the wrapped value directly to a BinaryOperatorAggregate channel.

Class

LangGraphDeprecationWarning

A LangGraph specific deprecation warning.

Class

LangGraphDeprecatedSinceV05

A specific LangGraphDeprecationWarning subclass defining functionality deprecated since LangGraph v0.5.0

Class

LangGraphDeprecatedSinceV10

A specific LangGraphDeprecationWarning subclass defining functionality deprecated since LangGraph v1.0.0

Class

Topic

A configurable PubSub Topic.

Class

BaseChannel

Base class for all channels.

Class

LastValue

Stores the last value received, can receive at most one value per step.

Class

LastValueAfterFinish

Stores the last value received, but only made available after finish().

Class

EphemeralValue

Stores the value received in the step immediately preceding, clears after.

Class

AnyValue

Stores the last value received, assumes that if multiple values are

Class

NamedBarrierValue

A channel that waits until all named values are received before making the value available.

Class

NamedBarrierValueAfterFinish

A channel that waits until all named values are received before making the value ready to be made available. It is only made available after finish() is called.

Class

BinaryOperatorAggregate

Stores the result of applying a binary operator to the current value and each new value.

Class

UntrackedValue

Stores the last value received, never checkpointed.

Class

StateGraph

A graph whose nodes communicate by reading and writing to a shared state.

Class

CompiledStateGraph

Class

UIMessage

A message type for UI updates in LangGraph.

Class

RemoveUIMessage

A message type for removing UI components in LangGraph.

Class

StateNodeSpec

Class

MessagesState

Class

BranchSpec

Class

entrypoint

Define a LangGraph workflow using the entrypoint decorator.

Class

final

A primitive that can be returned from an entrypoint.

Class

FunctionNonLocals

Get the nonlocal variables accessed of a function.

Class

NonLocals

Get nonlocal variables accessed.

Class

SyncAsyncFuture

Class

NodeBuilder

Class

Pregel

Pregel manages the runtime behavior for LangGraph applications.

Class

ChannelWriteEntry

Class

ChannelWriteTupleEntry

Class

ChannelWrite

Implements the logic for sending writes to CONFIG_KEY_SEND.

Class

StreamMessagesHandler

A callback handler that implements stream_mode=messages.

Class

TaskPayload

Class

TaskResultPayload

Class

CheckpointTask

Class

CheckpointPayload

Class

PregelProtocol

Class

StreamProtocol

Class

PregelLoop

Class

SyncPregelLoop

Class

AsyncPregelLoop

Class

ChannelRead

Implements the logic for reading state from CONFIG_KEY_READ.

Class

PregelNode

A node in a Pregel graph. This won't be invoked as a runnable by the graph

Class

Submit

Class

BackgroundExecutor

A context manager that runs sync tasks in the background.

Class

AsyncBackgroundExecutor

A context manager that runs async tasks in the background.

Class

RemoteException

Exception raised when an error occurs in the remote graph.

Class

RemoteGraph

The RemoteGraph class is a client implementation for calling remote

Class

WritesProtocol

Protocol for objects containing writes to be applied to checkpoint.

Class

PregelTaskWrites

Simplest implementation of WritesProtocol, for usage with writes that

Class

Call

Class

LazyAtomicCounter

Class

Edge

Class

TriggerEdge

Class

FuturesDict

Class

PregelRunner

Responsible for executing a set of Pregel tasks concurrently, committing

Class

IsLastStepManager

Class

RemainingStepsManager

Class

ManagedValue

Class

PregelScratchpad

Class

AsyncQueue

Async unbounded FIFO queue with a wait() method.

Class

Semaphore

Semaphore subclass with a wait() method.

Class

SyncQueue

Unbounded FIFO queue with a wait() method.

Class

TypedDictLikeV1

Protocol to represent types that behave like TypedDicts

Class

TypedDictLikeV2

Protocol to represent types that behave like TypedDicts

Class

DataclassLike

Protocol to represent types that behave like dataclasses.

Class

DeprecatedKwargs

TypedDict to use for extra keyword arguments, enabling type checking warnings for deprecated arguments.

Class

StrEnum

A string enum.

Class

RunnableCallable

A much simpler version of RunnableLambda that requires sync and async functions.

Class

RunnableSeq

Sequence of Runnable, where the output of each is the input of the next.

Class

NodeInterrupt

deprecated

Raised by a node to interrupt execution.

Class

MessageGraph

deprecated

A StateGraph where every node receives a list of messages as input and returns one or more messages as output.

Functions

Function

create_error_message

Function

get_runtime

Get the runtime for the current graph run.

Function

get_config

Function

get_store

Access LangGraph store from inside a graph node or entrypoint task at runtime.

Function

get_stream_writer

Access LangGraph StreamWriter from inside a graph node or entrypoint task at runtime.

Function

ensure_valid_checkpointer

Function

interrupt

Interrupt the graph with a resumable exception from within a node.

Function

push_ui_message

Push a new UI message to update the UI state.

Function

delete_ui_message

Delete a UI message by ID from the UI state.

Function

ui_message_reducer

Merge two lists of UI messages, supporting removing UI messages.

Function

add_messages

Merges two lists of messages, updating existing messages by ID.

Function

push_message

Write a message manually to the messages / messages-tuple stream mode.

Function

task

Define a LangGraph task using the task decorator.

Function

get_new_channel_versions

Get subset of current_versions that are newer than previous_versions.

Function

find_subgraph_pregel

Function

get_function_nonlocals

Get the nonlocal variables accessed by a function.

Function

is_xxh3_128_hexdigest

Check if the given string matches the format of xxh3_128_hexdigest.

Function

identifier

Return the module and name of an object.

Function

get_runnable_for_entrypoint

Function

get_runnable_for_task

Function

call

Function

empty_checkpoint

Function

create_checkpoint

Create a checkpoint for the given channels.

Function

channels_from_checkpoint

Get channels from a checkpoint.

Function

copy_checkpoint

Function

map_debug_tasks

Produce "task" events for stream_mode=debug.

Function

is_multiple_channel_write

Return True if the payload already wraps multiple writes from the same channel.

Function

map_task_result_writes

Folds task writes into a result dict and aggregates multiple writes to the same channel.

Function

map_debug_task_results

Produce "task_result" events for stream_mode=debug.

Function

rm_pregel_keys

Remove pregel-specific keys from the config.

Function

map_debug_checkpoint

Produce "checkpoint" events for stream_mode=debug.

Function

tasks_w_writes

Apply writes / subgraph states to tasks to be returned in a StateSnapshot.

Function

get_colored_text

Get colored text.

Function

get_bolded_text

Get bolded text.

Function

validate_graph

Function

validate_keys

Function

DuplexStream

Function

gated

A coroutine that waits for a semaphore before running another coroutine.

Function

next_tick

A function that yields control to other threads before running another function.

Function

read_channel

Function

read_channels

Function

map_command

Map input chunk to a sequence of pending writes in the form (channel, value).

Function

map_input

Map input chunk to a sequence of pending writes in the form (channel, value).

Function

map_output_values

Map pending writes (a sequence of tuples (channel, value)) to output chunk.

Function

map_output_updates

Map pending writes (a sequence of tuples (channel, value)) to output chunk.

Function

run_with_retry

Run a task with retries.

Function

arun_with_retry

Run a task asynchronously with retries.

Function

should_interrupt

Check if the graph should be interrupted based on current state.

Function

local_read

Function injected under CONFIG_KEY_READ in task config, to read current state.

Function

increment

Default channel versioning function, increments the current int version.

Function

apply_writes

Apply writes from a set of tasks (usually the tasks from a Pregel step)

Function

prepare_next_tasks

Prepare the set of tasks that will make up the next Pregel step.

Function

prepare_single_task

Prepares a single task for the next Pregel step, given a task path, which

Function

prepare_push_task_functional

Prepare a push task with an attached caller. Used for the functional API.

Function

prepare_push_task_send

Function

checkpoint_null_version

Get the null version for the checkpoint, if available.

Function

task_path_str

Generate a string representation of the task path.

Function

sanitize_untracked_values_in_send

Pop any values belonging to UntrackedValue channels in Send.arg for safe checkpointing.

Function

draw_graph

Get the graph for this Pregel instance.

Function

add_edge

Add an edge to the graph.

Function

is_managed_value

Function

default_cache_key

Default cache key function that uses the arguments and keyword arguments to generate a hashable key.

Function

get_field_default

Determine the default value for a field in a state schema.

Function

get_enhanced_type_hints

Attempt to extract default values and descriptions from provided type, used for config schema.

Function

get_update_as_tuples

Get Pydantic state update as a list of (key, value) tuples.

Function

get_cached_annotated_keys

Return cached annotated keys for a Python class.

Function

recast_checkpoint_ns

Remove task IDs from checkpoint namespace.

Function

patch_configurable

Function

patch_checkpoint_map

Function

merge_configs

Merge multiple configs into one.

Function

patch_config

Patch a config with new values.

Function

get_callback_manager_for_config

Get a callback manager for a config.

Function

get_async_callback_manager_for_config

Get an async callback manager for a config.

Function

ensure_config

Return a config with all keys, merging any provided configs.

Function

chain_future

Function

run_coroutine_threadsafe

Submit a coroutine object to a given event loop.

Function

apply_checkpointer_allowlist

Function

curated_core_allowlist

Function

build_serde_allowlist

Function

collect_allowlist_from_schemas

Function

set_config_context

Set the child Runnable config + tracing context.

Function

is_async_callable

Check if a function is async.

Function

is_async_generator

Check if a function is an async generator.

Function

coerce_to_runnable

Coerce a runnable-like object into a Runnable.

Function

get_fields

Get the field names of a Pydantic model.

Function

create_model

Create a pydantic model with the given field definitions.

Function

is_supported_by_pydantic

Check if a given "complex" type is supported by pydantic.

Function

default_retry_on

Modules

Module

langgraph

Module

version

Exports package version.

Module

errors

Module

runtime

Module

config

Module

typing

Module

types

Module

warnings

LangGraph specific warnings.

Module

constants

Module

channels

Module

topic

Module

base

Module

last_value

Module

ephemeral_value

Module

any_value

Module

named_barrier_value

Module

binop

Module

untracked_value

Module

graph

Module

state

Module

ui

Module

message

Module

utils

Legacy utilities module, to be removed in v1.

Module

config

Backwards compat imports for config utilities, to be removed in v1.

Module

runnable

Backwards compat imports for runnable utilities, to be removed in v1.

Module

func

Module

pregel

Module

main

Module

debug

Module

protocol

Module

types

Re-export types moved to langgraph.types

Module

remote

Module

managed

Module

is_last_step

Module

base

Types

Type

Checkpointer

Type of the checkpointer to use for a subgraph.

Type

AnyUIMessage

Type

StateNode

Type

Messages

Type

AnyFuture

Type

StateLike

Type alias for state-like types.

Type

RunnableLike

Menu

Language
Theme