temporalio.contrib.langgraph

package documentation

(source)

LangGraph plugin for Temporal SDK.

Warning

This package is experimental and may change in future versions. Use with caution in production environments.

This plugin runs LangGraph nodes and tasks as Temporal Activities, giving your AI agent workflows durable execution, automatic retries, and timeouts. It supports both the LangGraph Graph API (StateGraph) and Functional API (@entrypoint / @task).

Module	`_activity`	Activity wrappers for executing LangGraph nodes and tasks.
Module	`_interceptor`	Workflow interceptor that scopes LangGraph graphs/entrypoints to the workflow run.
Module	`_langgraph_config`	LangGraph configuration management for Temporal workflows.
Module	`_plugin`	LangGraph plugin for running LangGraph nodes and tasks as Temporal activities.
Module	`_task_cache`	Task result cache for continue-as-new support.

From __init__.py:

Class	`LangGraphPlugin`	LangGraph plugin for Temporal SDK.
Function	`cache`	Return the task result cache as a serializable dict.
Function	`entrypoint`	Retrieve a registered entrypoint by name.
Function	`graph`	Retrieve a registered graph by name.

def cache() -> dict[str, Any] | None: (source) ¶

Return the task result cache as a serializable dict.

Returns a dict suitable for passing to entrypoint(name, cache=...) to restore cached task results across continue-as-new boundaries. Returns None if the cache is empty.

def entrypoint(name: str, cache: dict[str, Any] | None = None) -> Pregel[Any, Any, Any, Any]: (source) ¶

Retrieve a registered entrypoint by name.

Parameters
name:`str`	Entrypoint name as registered with Plugin.
cache:`dict[str, Any] \| None`	Optional task result cache from a previous cache() call. Restores cached results so previously-completed tasks are not re-executed after continue-as-new.

Returns
`Pregel[Any, Any, Any, Any]`	Undocumented

def graph(name: str, cache: dict[str, Any] | None = None) -> StateGraph[Any, Any, Any, Any]: (source) ¶

Retrieve a registered graph by name.

Parameters
name:`str`	Graph name as registered with LangGraphPlugin.
cache:`dict[str, Any] \| None`	Optional task result cache from a previous cache() call. Restores cached results so previously-completed nodes are not re-executed after continue-as-new.

Returns
`StateGraph[Any, Any, Any, Any]`	Undocumented