Full config
Configurable
Bases: TypedDict
The raw config that is passed to langgraph client.
Note: ONLY use primitive types otherwise they don't show up in langgraph studio Note: This should not be used directly by nodes. They should use FullConfig.from_run_config(config)
assistant_name
instance-attribute
assistant_type
instance-attribute
assistant_version
instance-attribute
conversation_uid
instance-attribute
conversation_version
instance-attribute
running_as_subgraph
instance-attribute
use_latest_assistant
instance-attribute
user_id
instance-attribute
FullConfig
Bases: BaseModel
Config that is available to nodes at runtime.
Generally consider to be fixed for the duration of a thread, although technically new config can be passed when resuming.
Note: This validated version adds the dependency injections and additional methods for easier access to complex config data.
This is what nodes should use
assistant_config_id
instance-attribute
di
instance-attribute
main_conv_id
instance-attribute
model_config = ConfigDict(extra='ignore', arbitrary_types_allowed=True)
class-attribute
instance-attribute
running_as_subgraph = False
class-attribute
instance-attribute
user_pk_id
instance-attribute
from_run_config(config)
classmethod
Ensure that required config exists, and add the dependency injections.
Note: The Configurable type is not enforced at runtime, so have to do all validation here. Note: This must be re-run in every node. This is also necessary in-case the graph is resumed.
get_agent()
async
get_assistant()
async
get_chat_model()
async
get_options()
async
get_sub_assistant_infos()
async
get_user()
async
RequiresConfigError
Bases: LanggraphError
Error raised when a node requires a config but it is not available.