AgentExecutorIterator#

class langchain.agents.agent_iterator.AgentExecutorIterator(agent_executor: AgentExecutor, inputs: Any, callbacks: Callbacks = None, *, tags: list[str] | None = None, metadata: Dict[str, Any] | None = None, run_name: str | None = None, run_id: UUID | None = None, include_run_info: bool = False, yield_actions: bool = False)[source]#

Iterator for AgentExecutor.

Initialize the AgentExecutorIterator with the given AgentExecutor, inputs, and optional callbacks.

Parameters:

agent_executor (AgentExecutor) – The AgentExecutor to iterate over.
inputs (Any) – The inputs to the AgentExecutor.
callbacks (Callbacks, optional) – The callbacks to use during iteration. Defaults to None.
tags (Optional[list[str]], optional) – The tags to use during iteration. Defaults to None.
metadata (Optional[Dict[str, Any]], optional) – The metadata to use during iteration. Defaults to None.
run_name (Optional[str], optional) – The name of the run. Defaults to None.
run_id (Optional[UUID], optional) – The ID of the run. Defaults to None.
include_run_info (bool, optional) – Whether to include run info in the output. Defaults to False.
yield_actions (bool, optional) – Whether to yield actions as they are generated. Defaults to False.

Attributes

`agent_executor`	The AgentExecutor to iterate over.
`color_mapping`	A mapping of tool names to colors.
`inputs`	The inputs to the AgentExecutor.
`name_to_tool_map`	A mapping of tool names to tools.

Methods

`__init__`(agent_executor, inputs[, ...])	Initialize the AgentExecutorIterator with the given AgentExecutor, inputs, and optional callbacks.
`make_final_outputs`(outputs, run_manager)
`reset`()	Reset the iterator to its initial state, clearing intermediate steps, iterations, and time elapsed.
`update_iterations`()	Increment the number of iterations and update the time elapsed.

__init__(agent_executor: AgentExecutor, inputs: Any, callbacks: Callbacks = None, *, tags: list[str] | None = None, metadata: Dict[str, Any] | None = None, run_name: str | None = None, run_id: UUID | None = None, include_run_info: bool = False, yield_actions: bool = False)[source]#

Initialize the AgentExecutorIterator with the given AgentExecutor, inputs, and optional callbacks.

Parameters:

agent_executor (AgentExecutor) – The AgentExecutor to iterate over.
inputs (Any) – The inputs to the AgentExecutor.
callbacks (Callbacks, optional) – The callbacks to use during iteration. Defaults to None.
tags (Optional[list[str]], optional) – The tags to use during iteration. Defaults to None.
metadata (Optional[Dict[str, Any]], optional) – The metadata to use during iteration. Defaults to None.
run_name (Optional[str], optional) – The name of the run. Defaults to None.
run_id (Optional[UUID], optional) – The ID of the run. Defaults to None.
include_run_info (bool, optional) – Whether to include run info in the output. Defaults to False.
yield_actions (bool, optional) – Whether to yield actions as they are generated. Defaults to False.

make_final_outputs(outputs: Dict[str, Any], run_manager: CallbackManagerForChainRun | AsyncCallbackManagerForChainRun) → AddableDict[source]#

Parameters:

outputs (Dict[str, Any])
run_manager (CallbackManagerForChainRun | AsyncCallbackManagerForChainRun)

Return type:

AddableDict

reset() → None[source]#

Reset the iterator to its initial state, clearing intermediate steps, iterations, and time elapsed.

Return type:: None

update_iterations() → None[source]#

Increment the number of iterations and update the time elapsed.

Return type:: None