Constructs the agent scratchpad based on the agent steps. It returns an array of base messages representing the thoughts of the agent.
The agent steps to construct the scratchpad from.
An array of base messages representing the thoughts of the agent.
Decide what to do given some input.
Steps the LLM has taken so far, along with observations from each.
User inputs.
Optional
callbackManager: anyCallback manager to use for this call.
Action specifying what tool to use.
Return response when agent has been stopped due to max iterations
Optional
callbackManager: anyStatic
createCreate prompt in the style of the ChatConversationAgent.
List of tools the agent will have access to, used to format the prompt.
Optional
args: ChatConversationalCreatePromptArgsArguments to create the prompt with.
Static
deserializeLoad an agent from a json-like object describing it.
Static
fromLLMAndCreates an instance of the ChatConversationalAgent class from a BaseLanguageModel and a set of tools. It takes optional arguments to customize the agent.
The BaseLanguageModel to create the agent from.
The set of tools to create the agent from.
Optional
args: ChatConversationalCreatePromptArgs & AgentArgsOptional arguments to customize the agent.
An instance of the ChatConversationalAgent class.
Static
getReturns the default output parser for the ChatConversationalAgent class. It takes optional fields as arguments to customize the output parser.
Optional
fields: OutputParserArgs & { Optional fields to customize the output parser.
The default output parser for the ChatConversationalAgent class.
Static
validateGenerated using TypeDoc
Agent for the MRKL chain.
⚠️ Deprecated ⚠️
Use the createStructuredChatAgent method instead.
This feature is deprecated and will be removed in the future.
It is not recommended for use.