Class ZeroShotAgent

Agent for the MRKL chain.

Example


const agent = new ZeroShotAgent({
  llmChain: new LLMChain({
    llm: new ChatOpenAI({ temperature: 0 }),
    prompt: ZeroShotAgent.createPrompt([new SerpAPI(), new Calculator()], {
      prefix: `Answer the following questions as best you can, but speaking as a pirate might speak. You have access to the following tools:`,
      suffix: `Begin! Remember to speak as a pirate when giving your final answer. Use lots of "Args"
Question: {input}
{agent_scratchpad}`,
      inputVariables: ["input", "agent_scratchpad"],
    }),
  }),
  allowedTools: ["search", "calculator"],
});

const result = await agent.invoke({
  input: `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`,
});

Hierarchy

Agent
- ZeroShotAgent

Index

Constructors

constructor

new ZeroShotAgent(input): ZeroShotAgent
Parameters
- input: ZeroShotAgentInput
Returns ZeroShotAgent
Overrides Agent.constructor
- Defined in langchain/src/agents/mrkl/index.ts:69

Properties

ToolType

ToolType: Tool

llmChain

llmChain: LLMChain<string, LLMType>

outputParser

outputParser: undefined | AgentActionOutputParser

Accessors

allowedTools

get allowedTools(): undefined | string[]
Returns undefined | string[]
Inherited from Agent.allowedTools
- Defined in langchain/src/agents/agent.ts:319

inputKeys

get inputKeys(): string[]
Returns string[]
Inherited from Agent.inputKeys
- Defined in langchain/src/agents/agent.ts:323

returnValues

get returnValues(): string[]
Returns string[]
Inherited from Agent.returnValues
- Defined in langchain/src/agents/agent.ts:51

Methods

constructScratchPad

constructScratchPad(steps): Promise<string | BaseMessage[]>
Construct a scratchpad to let the agent continue its thought process
Parameters
- steps: AgentStep[]
Returns Promise<string | BaseMessage[]>
Inherited from Agent.constructScratchPad
- Defined in langchain/src/agents/agent.ts:404

finishToolName

finishToolName(): string
Name of tool to use to terminate the chain.

Returns string
Inherited from Agent.finishToolName
- Defined in langchain/src/agents/agent.ts:397

llmPrefix

llmPrefix(): string
Prefix to append the LLM call with.

Returns string
Overrides Agent.llmPrefix
- Defined in langchain/src/agents/mrkl/index.ts:83

observationPrefix

observationPrefix(): string
Prefix to append the observation with.

Returns string
Overrides Agent.observationPrefix
- Defined in langchain/src/agents/mrkl/index.ts:79

plan

plan(steps, inputs, callbackManager?): Promise<AgentAction | AgentFinish>
Decide what to do given some input.
Parameters
- steps: AgentStep[]
  
  Steps the LLM has taken so far, along with observations from each.
- inputs: ChainValues
  
  User inputs.
- Optional callbackManager: CallbackManager
  
  Callback manager to use for this call.
Returns Promise<AgentAction | AgentFinish>
Action specifying what tool to use.
Inherited from Agent.plan
- Defined in langchain/src/agents/agent.ts:451

prepareForOutput

prepareForOutput(_returnValues, _steps): Promise<Record<string, any>>
Prepare the agent for output, if needed
Parameters
- _returnValues: Record<string, any>
- _steps: AgentStep[]
Returns Promise<Record<string, any>>
Inherited from Agent.prepareForOutput
- Defined in langchain/src/agents/agent.ts:93

returnStoppedResponse

returnStoppedResponse(earlyStoppingMethod, steps, inputs, callbackManager?): Promise<AgentFinish>
Return response when agent has been stopped due to max iterations
Parameters
- earlyStoppingMethod: StoppingMethod
- steps: AgentStep[]
- inputs: ChainValues
- Optional callbackManager: CallbackManager
Returns Promise<AgentFinish>
Inherited from Agent.returnStoppedResponse
- Defined in langchain/src/agents/agent.ts:462

toJSON

toJSON(): Serialized
Returns Serialized
Inherited from Agent.toJSON
- Defined in langchain-core/dist/load/serializable.d.ts:64

toJSONNotImplemented

toJSONNotImplemented(): SerializedNotImplemented
Returns SerializedNotImplemented
Inherited from Agent.toJSONNotImplemented
- Defined in langchain-core/dist/load/serializable.d.ts:65

`Static` createPrompt

createPrompt(tools, args?): PromptTemplate<any, any>
Create prompt in the style of the zero shot agent.
Parameters
- tools: Tool[]
  
  List of tools the agent will have access to, used to format the prompt.
- Optional args: ZeroShotCreatePromptArgs
  
  Arguments to create the prompt with.
Returns PromptTemplate<any, any>
Overrides Agent.createPrompt
- Defined in langchain/src/agents/mrkl/index.ts:120

`Static` deserialize

deserialize(data): Promise<ZeroShotAgent>
Load an agent from a json-like object describing it.
Parameters
- data: Object
Returns Promise<ZeroShotAgent>
Overrides Agent.deserialize
- Defined in langchain/src/agents/mrkl/index.ts:175

`Static` fromLLMAndTools

fromLLMAndTools(llm, tools, args?): ZeroShotAgent
Creates a ZeroShotAgent from a Large Language Model and a set of tools.
Parameters
- llm: BaseLanguageModel<any, BaseLanguageModelCallOptions>
  
  The Large Language Model to use.
- tools: Tool[]
  
  The tools for the agent to use.
- Optional args: ZeroShotCreatePromptArgs & AgentArgs
  
  Optional arguments for creating the agent.
Returns ZeroShotAgent
A new instance of ZeroShotAgent.
Overrides Agent.fromLLMAndTools
- Defined in langchain/src/agents/mrkl/index.ts:153

`Static` getDefaultOutputParser

getDefaultOutputParser(fields?): ZeroShotAgentOutputParser
Returns the default output parser for the ZeroShotAgent.
Parameters
- Optional fields: OutputParserArgs
  
  Optional arguments for the output parser.
Returns ZeroShotAgentOutputParser
An instance of ZeroShotAgentOutputParser.
Overrides Agent.getDefaultOutputParser
- Defined in langchain/src/agents/mrkl/index.ts:92

`Static` validateTools

validateTools(tools): void
Validates the tools for the ZeroShotAgent. Throws an error if any tool does not have a description.
Parameters
- tools: Tool[]
  
  List of tools to validate.
Returns void
Overrides Agent.validateTools
- Defined in langchain/src/agents/mrkl/index.ts:101

Class ZeroShotAgent

Example

Hierarchy

Index

Constructors

Properties

Accessors

Methods

Constructors

constructor

Parameters

input: ZeroShotAgentInput

Returns ZeroShotAgent

Properties

ToolType

llmChain

outputParser

Accessors

allowedTools

Returns undefined | string[]

inputKeys

Returns string[]

returnValues

Returns string[]

Methods

constructScratchPad

Parameters

steps: AgentStep[]

Returns Promise<string | BaseMessage[]>

finishToolName

Returns string

llmPrefix

Returns string

observationPrefix

Returns string

plan

Parameters

steps: AgentStep[]

inputs: ChainValues

Optional callbackManager: CallbackManager

Returns Promise<AgentAction | AgentFinish>

prepareForOutput

Parameters

_returnValues: Record<string, any>

_steps: AgentStep[]

Returns Promise<Record<string, any>>

returnStoppedResponse

Parameters

earlyStoppingMethod: StoppingMethod

steps: AgentStep[]

inputs: ChainValues

Optional callbackManager: CallbackManager

Returns Promise<AgentFinish>

toJSON

Returns Serialized

toJSONNotImplemented

Returns SerializedNotImplemented

Static createPrompt

Parameters

tools: Tool[]

Optional args: ZeroShotCreatePromptArgs

Returns PromptTemplate<any, any>

Static deserialize

Parameters

data: Object

Returns Promise<ZeroShotAgent>

Static fromLLMAndTools

Parameters

llm: BaseLanguageModel<any, BaseLanguageModelCallOptions>

tools: Tool[]

Optional args: ZeroShotCreatePromptArgs & AgentArgs

Returns ZeroShotAgent

Static getDefaultOutputParser

Parameters

Optional fields: OutputParserArgs

Returns ZeroShotAgentOutputParser

Static validateTools

Parameters

tools: Tool[]

Returns void

`Optional` callbackManager: CallbackManager

`Optional` callbackManager: CallbackManager

`Static` createPrompt

`Optional` args: ZeroShotCreatePromptArgs

`Static` deserialize

`Static` fromLLMAndTools

`Optional` args: ZeroShotCreatePromptArgs & AgentArgs

`Static` getDefaultOutputParser

`Optional` fields: OutputParserArgs

`Static` validateTools