StructuredChatAgent#

class langchain.agents.structured_chat.base.StructuredChatAgent[source]#

Bases: Agent

Deprecated since version 0.1.0: Use create_structured_chat_agent() instead.

Structured Chat Agent.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

param allowed_tools: List[str] | None = None#

Allowed tools for the agent. If None, all tools are allowed.

param llm_chain: LLMChain [Required]#

LLMChain to use for agent.

param output_parser: AgentOutputParser [Optional]#

Output parser for the agent.

async aplan(intermediate_steps: List[Tuple[AgentAction, str]], callbacks: list[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → AgentAction | AgentFinish#

Async given input, decided what to do.

Parameters:
Returns:

Action specifying what tool to use.

Return type:

AgentAction | AgentFinish

classmethod create_prompt(tools: Sequence[BaseTool], prefix: str = 'Respond to the human as helpfully and accurately as possible. You have access to the following tools:', suffix: str = 'Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation:.\nThought:', human_message_template: str = '{input}\n\n{agent_scratchpad}', format_instructions: str = 'Use a json blob to specify a tool by providing an action key (tool name) and an action_input key (tool input).\n\nValid "action" values: "Final Answer" or {tool_names}\n\nProvide only ONE action per $JSON_BLOB, as shown:\n\n```\n{{{{\n  "action": $TOOL_NAME,\n  "action_input": $INPUT\n}}}}\n```\n\nFollow this format:\n\nQuestion: input question to answer\nThought: consider previous and subsequent steps\nAction:\n```\n$JSON_BLOB\n```\nObservation: action result\n... (repeat Thought/Action/Observation N times)\nThought: I know what to respond\nAction:\n```\n{{{{\n  "action": "Final Answer",\n  "action_input": "Final response to human"\n}}}}\n```', input_variables: List[str] | None = None, memory_prompts: List[BasePromptTemplate] | None = None) → BasePromptTemplate[source]#

Create a prompt for this class.

Parameters:

  • tools (Sequence[BaseTool]) – Tools to use.

  • prefix (str)

  • suffix (str)

  • human_message_template (str)

  • format_instructions (str)

  • input_variables (List[str] | None)

  • memory_prompts (List[BasePromptTemplate] | None)

Returns:

Prompt template.

Return type:

BasePromptTemplate

classmethod from_llm_and_tools(llm: BaseLanguageModel, tools: Sequence[BaseTool], callback_manager: BaseCallbackManager | None = None, output_parser: AgentOutputParser | None = None, prefix: str = 'Respond to the human as helpfully and accurately as possible. You have access to the following tools:', suffix: str = 'Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation:.\nThought:', human_message_template: str = '{input}\n\n{agent_scratchpad}', format_instructions: str = 'Use a json blob to specify a tool by providing an action key (tool name) and an action_input key (tool input).\n\nValid "action" values: "Final Answer" or {tool_names}\n\nProvide only ONE action per $JSON_BLOB, as shown:\n\n```\n{{{{\n  "action": $TOOL_NAME,\n  "action_input": $INPUT\n}}}}\n```\n\nFollow this format:\n\nQuestion: input question to answer\nThought: consider previous and subsequent steps\nAction:\n```\n$JSON_BLOB\n```\nObservation: action result\n... (repeat Thought/Action/Observation N times)\nThought: I know what to respond\nAction:\n```\n{{{{\n  "action": "Final Answer",\n  "action_input": "Final response to human"\n}}}}\n```', input_variables: List[str] | None = None, memory_prompts: List[BasePromptTemplate] | None = None, **kwargs: Any) → Agent[source]#

Construct an agent from an LLM and tools.

Parameters:
Return type:

Agent

get_allowed_tools() → List[str] | None#

Get allowed tools.

Return type:

List[str] | None

get_full_inputs(intermediate_steps: List[Tuple[AgentAction, str]], **kwargs: Any) → Dict[str, Any]#

Create the full inputs for the LLMChain from intermediate steps.

Parameters:

  • intermediate_steps (List[Tuple[AgentAction, str]]) – Steps the LLM has taken to date, along with observations.

  • **kwargs (Any) – User inputs.

Returns:

Full inputs for the LLMChain.

Return type:

Dict[str, Any]

plan(intermediate_steps: List[Tuple[AgentAction, str]], callbacks: list[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → AgentAction | AgentFinish#

Given input, decided what to do.

Parameters:
Returns:

Action specifying what tool to use.

Return type:

AgentAction | AgentFinish

return_stopped_response(early_stopping_method: str, intermediate_steps: List[Tuple[AgentAction, str]], **kwargs: Any) → AgentFinish#

Return response when agent has been stopped due to max iterations.

Parameters:

  • early_stopping_method (str) – Method to use for early stopping.

  • intermediate_steps (List[Tuple[AgentAction, str]]) – Steps the LLM has taken to date, along with observations.

  • **kwargs (Any) – User inputs.

Returns:

Agent finish object.

Return type:

AgentFinish

Raises:

ValueError – If early_stopping_method is not in [‘force’, ‘generate’].

save(file_path: Path | str) → None#

Save the agent.

Parameters:

file_path (Path | str) – Path to file to save the agent to.

Return type:

None

Example: .. code-block:: python

# If working with agent executor agent.agent.save(file_path=”path/agent.yaml”)

tool_run_logging_kwargs() → Dict#

Return logging kwargs for tool run.

Return type:

Dict

property llm_prefix: str#

Prefix to append the llm call with.

property observation_prefix: str#

Prefix to append the observation with.

property return_values: List[str]#

Return values of the agent.

Examples using StructuredChatAgent