Component

Invoker-based component templates used across Gen AI applications.

GenericLMComponent(lm_invoker, fallback_lms=None)

Bases: LMComponent

A general-purpose LM component.

Examples:

Basic usage

component = GenericLMComponent.from_config(model_id="openai/gpt-5.4-nano")
response = await component.run(query="What is the capital of France?")

With custom prompt templates

component = GenericLMComponent.from_config(
    model_id="openai/gpt-5.4-nano",
    system_template="Talk like a {role}.",
)
response = await component.run(query="What is the capital of France?", role="pirate")

Attributes:

Name	Type	Description
prompt_vars	set[str]	Required prompt variables.
default_system_template	str	Default system template.
default_user_template	str	Default user template.

invoke(query, history=None, extra_contents=None, hyperparameters=None, event_emitter=None, max_calls=1, **prompt_kwargs) async

Invokes the language model.

Parameters:

Name	Type	Description	Default
query	str	The query to be passed to the language model.	required
history	list[Message] | None	Conversation history to include in the prompt. Defaults to None.	None
extra_contents	list[MessageContent] | None	Extra user contents for prompt formatting. Defaults to None.	None
hyperparameters	dict[str, Any] | None	Hyperparameters passed to the language model invocation. Defaults to None.	None
event_emitter	EventEmitter | None	Event emitter for streaming outputs. Defaults to None.	None
max_calls	int	The max number of times the language model can be invoked. Defaults to 1.	1
**prompt_kwargs	Any	Additional prompt variables.	{}

Returns:

Name	Type	Description
LMOutput	LMOutput	The output of the language model.

LMComponent(lm_invoker, fallback_lms=None)

Bases: Component

A template component for LM-based components used across Gen AI applications.

Single-slot components keep the existing API: define class-level prompt fields and pass one invoker. Multi-slot components declare lm_slots and pass a mapping of slot names to invokers.

Examples:

Define a single-slot LM component class

class MyLMComponent(LMComponent):
    prompt_vars = {"context", "query"}
    default_system_template = "Answer the query based on the following context: {context}"
    default_user_template = "{query}"

Define a multi-slot LM component class

class MapReduceComponent(LMComponent):
    lm_slots = {
        "map": LMComponentSlot(
            prompt_vars={"context", "query"},
            default_system_template="Map the context for query: {query}",
            default_user_template="{context}",
        ),
        "reduce": LMComponentSlot(
            prompt_vars={"context", "query"},
            default_system_template="Reduce the mapped context for query: {query}",
            default_user_template="{context}",
        ),
    }

Init with LM invoker

component = MyLMComponent(lm_invoker=lm_invoker)
component = MapReduceComponent(lm_invoker={"map": map_invoker, "reduce": reduce_invoker})

Init with LM configuration

component = MyLMComponent.from_config(
    model_id="openai/gpt-5.4-nano",
    config={"response_schema": ResponseSchema},
)

Attributes:

Name	Type	Description
prompt_vars	set[str]	Required prompt variables for the implicit default slot.
default_system_template	str	Default system template for the implicit default slot.
default_user_template	str	Default user template for the implicit default slot.
lm_slots	dict[str, LMComponentSlot] | None	Explicit named slots for multi-LM components.
fallback_lms	list[BaseLMInvoker] | dict[str, list[BaseLMInvoker]]	Ordered fallback invokers.

Initializes LMComponent with one or more LM invokers.

Parameters:

Name	Type	Description	Default
lm_invoker	BaseLMInvoker | dict[str, BaseLMInvoker]	The language model invoker or slot mapping.	required
fallback_lms	list[BaseLMInvoker] | dict[str, list[BaseLMInvoker]] | None	Ordered fallback invokers. Lists map to the implicit default slot. Dicts map to explicit slots. Defaults to None.	None

Raises:

Type	Description
ValueError

fallback_lms property

Returns the fallback LMs.

Returns:

Type	Description
list[BaseLMInvoker] | dict[str, list[BaseLMInvoker]]	list[BaseLMInvoker] | dict[str, list[BaseLMInvoker]]: The fallback LMs.

lm_invoker property

Returns the default LM invoker.

Returns:

Type	Description
BaseLMInvoker | dict[str, BaseLMInvoker]	BaseLMInvoker | dict[str, BaseLMInvoker]: The default LM invoker.

from_config(model_id, credentials=None, config=None, system_template=None, user_template=None, prompt_builder_kwargs=None, fallback_lms=None, **kwargs) classmethod

Builds an LM invoker from config and initializes the component.

This method can be used to initialize the component with a model id and its configuration. Currently only supported for single-slot components.

Parameters:

Name	Type	Description	Default
model_id	str | ModelId	The model ID.	required
credentials	str | dict[str, Any] | None	Model credentials. Defaults to None.	None
config	dict[str, Any] | None	Model configuration. Defaults to None.	None
system_template	str | None	System template. Defaults to None.	None
user_template	str | None	User template. Defaults to None.	None
prompt_builder_kwargs	dict[str, Any] | None	Prompt builder kwargs. Defaults to None.	None
fallback_lms	list[BaseLMInvoker] | None	Ordered fallback invokers. Defaults to None.	None
**kwargs	Any	Additional constructor kwargs for subclasses.	{}

Returns:

Name	Type	Description
Self	Self	A new LMComponent instance.

Raises:

Type	Description
ValueError	If prompt_builder_kwargs is provided without system_template or user_template.

LMComponentSlot

Bases: BaseModel

Prompt contract for one named LM slot in an LMComponent.

Attributes:

Name	Type	Description
prompt_vars	set[str]	Required prompt variables for the slot.
default_system_template	str	Default system template for the slot.
default_user_template	str	Default user template for the slot.