Class ChatAgent

Agent for the MRKL chain.

Hierarchy (View Summary)

Agent
- ChatAgent

Constructors

constructor

new ChatAgent(input: ChatAgentInput): ChatAgent
Parameters
- input: ChatAgentInput
Returns ChatAgent
Overrides Agent.constructor
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:53

Properties

lc_namespace

lc_namespace: string[] = ...

llmChain

llmChain: LLMChain

outputParser

outputParser: undefined | AgentActionOutputParser

ToolType

ToolType: ToolInterface

Accessors

allowedTools

get allowedTools(): undefined | string[]
Returns undefined | string[]
Inherited from Agent.allowedTools
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:488

inputKeys

get inputKeys(): string[]
Returns string[]
Inherited from Agent.inputKeys
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:492

returnValues

get returnValues(): string[]
Returns string[]
Inherited from Agent.returnValues
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:57

Methods

_agentActionType

_agentActionType(): string
Return the string type key uniquely identifying multi or single action agents.

Returns string
Inherited from Agent._agentActionType
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:113

_agentType

_agentType(): "chat-zero-shot-react-description"
Return the string type key uniquely identifying this class of agent.

Returns "chat-zero-shot-react-description"
Overrides Agent._agentType
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:59

_stop

_stop(): string[]
Returns string[]
Overrides Agent._stop
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:71

constructScratchPad

constructScratchPad(steps: AgentStep[]): Promise<string>
Constructs the agent's scratchpad, which is a string representation of the agent's previous steps.
Parameters
- steps: AgentStep[]
  Array of AgentStep instances representing the agent's previous steps.
Returns Promise<string>
Promise resolving to a string representing the agent's scratchpad.
Overrides Agent.constructScratchPad
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:106

finishToolName

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

Returns string
Inherited from Agent.finishToolName
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:566

llmPrefix

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

Returns string
Overrides Agent.llmPrefix
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:67

observationPrefix

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

Returns string
Overrides Agent.observationPrefix
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:63

plan

plan(
    steps: AgentStep[],
    inputs: ChainValues,
    callbackManager?: any,
): Promise<any>
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.
- OptionalcallbackManager: any
  Callback manager to use for this call.
Returns Promise<any>
Action specifying what tool to use.
Inherited from Agent.plan
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:620

prepareForOutput

prepareForOutput(
_returnValues: AgentFinish,
_steps: AgentStep[],
): Promise<AgentFinish>
Prepare the agent for output, if needed
Parameters
- _returnValues: AgentFinish
- _steps: AgentStep[]
Returns Promise<AgentFinish>
Inherited from Agent.prepareForOutput
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:99

returnStoppedResponse

returnStoppedResponse(
    earlyStoppingMethod: StoppingMethod,
    steps: AgentStep[],
    inputs: ChainValues,
    callbackManager?: any,
): Promise<AgentFinish>
Return response when agent has been stopped due to max iterations
Parameters
- earlyStoppingMethod: StoppingMethod
- steps: AgentStep[]
- inputs: ChainValues
- OptionalcallbackManager: any
Returns Promise<AgentFinish>
Inherited from Agent.returnStoppedResponse
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:631

`Static`createPrompt

createPrompt(tools: ToolInterface[], args?: ChatCreatePromptArgs): any
Create prompt in the style of the zero shot agent.
Parameters
- tools: ToolInterface[]
  List of tools the agent will have access to, used to format the prompt.
- Optionalargs: ChatCreatePromptArgs
  Arguments to create the prompt with.
  
  Interface for arguments used to create a chat prompt.
  - OptionalformatInstructions?: string
    Formattable string to use as the instructions template.
  - OptionalhumanMessageTemplate?: string
    String to use directly as the human message template.
  - OptionalinputVariables?: string[]
    List of input variables the final prompt will expect.
  - Optionalprefix?: string
    String to put before the list of tools.
  - Optionalsuffix?: string
    String to put after the list of tools.
Returns any
Overrides Agent.createPrompt
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:124

`Static`deserialize

deserialize(
data: SerializedZeroShotAgent & { llm?: any; tools?: ToolInterface[] },
): Promise<Agent>
Load an agent from a json-like object describing it.
Parameters
- data: SerializedZeroShotAgent & { llm?: any; tools?: ToolInterface[] }
Returns Promise<Agent>
Inherited from Agent.deserialize
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/agent.ts:673

`Static`fromLLMAndTools

fromLLMAndTools(
    llm: BaseLanguageModelInterface,
    tools: ToolInterface[],
    args?: ChatCreatePromptArgs & AgentArgs,
): ChatAgent
Creates a ChatAgent instance using a language model, tools, and optional arguments.
Parameters
- llm: BaseLanguageModelInterface
  BaseLanguageModelInterface instance to use in the agent.
- tools: ToolInterface[]
  Array of Tool instances to include in the agent.
- Optionalargs: ChatCreatePromptArgs & AgentArgs
  Optional arguments to customize the agent and prompt.
Returns ChatAgent
ChatAgent instance
Overrides Agent.fromLLMAndTools
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:152

`Static`getDefaultOutputParser

getDefaultOutputParser(_fields?: OutputParserArgs): ChatAgentOutputParser
Returns a default output parser for the ChatAgent.
Parameters
- Optional_fields: OutputParserArgs
  Optional OutputParserArgs to customize the output parser.
Returns ChatAgentOutputParser
ChatAgentOutputParser instance
Overrides Agent.getDefaultOutputParser
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:96

`Static`lc_name

lc_name(): string
Returns string
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:45

`Static`validateTools

validateTools(tools: ToolInterface[]): void
Validates that all tools have descriptions. Throws an error if a tool without a description is found.
Parameters
- tools: ToolInterface[]
  Array of Tool instances to validate.
Returns void
void
Overrides Agent.validateTools
- Defined in remotes/langchain-ai/langchainjs/main/libs/langchain-classic/src/agents/chat/index.ts:81

Class ChatAgent

Hierarchy (View Summary)

Index

Constructors

Properties

Accessors

Methods

Constructors

constructor

Parameters

Returns ChatAgent

Properties

lc_namespace

llmChain

outputParser

ToolType

Accessors

allowedTools

Returns undefined | string[]

inputKeys

Returns string[]

returnValues

Returns string[]

Methods

_agentActionType

Returns string

_agentType

Returns "chat-zero-shot-react-description"

_stop

Returns string[]

constructScratchPad

Parameters

Returns Promise<string>

finishToolName

Returns string

llmPrefix

Returns string

observationPrefix

Returns string

plan

Parameters

Returns Promise<any>

prepareForOutput

Parameters

Returns Promise<AgentFinish>

returnStoppedResponse

Parameters

Returns Promise<AgentFinish>

StaticcreatePrompt

Parameters

OptionalformatInstructions?: string

OptionalhumanMessageTemplate?: string

OptionalinputVariables?: string[]

Optionalprefix?: string

Optionalsuffix?: string

Returns any

Staticdeserialize

Parameters

Returns Promise<Agent>

StaticfromLLMAndTools

Parameters

Returns ChatAgent

StaticgetDefaultOutputParser

Parameters

Returns ChatAgentOutputParser

Staticlc_name

Returns string

StaticvalidateTools

Parameters

Returns void

Settings

On This Page

`Static`createPrompt

`Optional`formatInstructions?: string

`Optional`humanMessageTemplate?: string

`Optional`inputVariables?: string[]

`Optional`prefix?: string

`Optional`suffix?: string

`Static`deserialize

`Static`fromLLMAndTools

`Static`getDefaultOutputParser

`Static`lc_name

`Static`validateTools