Class: OpenAIAgent
Hierarchy
-
AgentRunner<OpenAI>↳
OpenAIAgent
Constructors
constructor
• new OpenAIAgent(params): OpenAIAgent
Parameters
| Name | Type |
|---|---|
params | OpenAIAgentParams |
Returns
Overrides
AgentRunner<OpenAI>.constructor
Defined in
packages/core/src/agent/openai.ts:36
Properties
#chatHistory
• Private #chatHistory: ChatMessage<ToolCallLLMMessageOptions>[]
Inherited from
AgentRunner.#chatHistory
Defined in
packages/core/src/agent/base.ts:230
#llm
• Private Readonly #llm: OpenAI
Inherited from
AgentRunner.#llm
Defined in
packages/core/src/agent/base.ts:225
#runner
• Private Readonly #runner: AgentWorker<OpenAI, {}, ToolCallLLMMessageOptions>
Inherited from
AgentRunner.#runner
Defined in
packages/core/src/agent/base.ts:231
#systemPrompt
• Private Readonly #systemPrompt: null | MessageContent = null
Inherited from
AgentRunner.#systemPrompt
Defined in
packages/core/src/agent/base.ts:229
#tools
• Private Readonly #tools: BaseToolWithCall[] | (query: MessageContent) => Promise<BaseToolWithCall[]>
Inherited from
AgentRunner.#tools
Defined in
packages/core/src/agent/base.ts:226
#verbose
• Private Readonly #verbose: boolean
Inherited from
AgentRunner.#verbose
Defined in
packages/core/src/agent/base.ts:232
createStore
• createStore: () => object = AgentRunner.defaultCreateStore
Type declaration
▸ (): object
Returns
object
Overrides
AgentRunner.createStore
Defined in
packages/core/src/agent/openai.ts:53
taskHandler
▪ Static taskHandler: TaskHandler<OpenAI>
Defined in
packages/core/src/agent/openai.ts:55
Accessors
chatHistory
• get chatHistory(): ChatMessage<AdditionalMessageOptions>[]
Returns
ChatMessage<AdditionalMessageOptions>[]
Inherited from
AgentRunner.chatHistory
Defined in
packages/core/src/agent/base.ts:259
llm
• get llm(): AI
Returns
AI
Inherited from
AgentRunner.llm
Defined in
packages/core/src/agent/base.ts:255
verbose
• get verbose(): boolean
Returns
boolean
Inherited from
AgentRunner.verbose
Defined in
packages/core/src/agent/base.ts:263
Methods
chat
▸ chat(params): Promise<AgentChatResponse<ToolCallLLMMessageOptions>>
Parameters
| Name | Type |
|---|---|
params | ChatEngineParamsNonStreaming |
Returns
Promise<AgentChatResponse<ToolCallLLMMessageOptions>>
Inherited from
AgentRunner.chat
Defined in
packages/core/src/agent/base.ts:329
▸ chat(params): Promise<ReadableStream<AgentStreamChatResponse<ToolCallLLMMessageOptions>>>
Parameters
| Name | Type |
|---|---|
params | ChatEngineParamsStreaming |
Returns
Promise<ReadableStream<AgentStreamChatResponse<ToolCallLLMMessageOptions>>>
Inherited from
AgentRunner.chat
Defined in
packages/core/src/agent/base.ts:332
createTask
▸ createTask(message, stream?, verbose?): ReadableStream<TaskStepOutput<OpenAI, {}, ToolCallLLMMessageOptions>>
Parameters
| Name | Type | Default value |
|---|---|---|
message | MessageContent | undefined |
stream | boolean | false |
verbose | undefined | boolean | undefined |
Returns
ReadableStream<TaskStepOutput<OpenAI, {}, ToolCallLLMMessageOptions>>
Inherited from
AgentRunner.createTask
Defined in
packages/core/src/agent/base.ts:290
getTools
▸ getTools(query): BaseToolWithCall[] | Promise<BaseToolWithCall[]>
Parameters
| Name | Type |
|---|---|
query | MessageContent |
Returns
BaseToolWithCall[] | Promise<BaseToolWithCall[]>
Inherited from
AgentRunner.getTools
Defined in
packages/core/src/agent/base.ts:271
reset
▸ reset(): void
Returns
void
Inherited from
AgentRunner.reset
Defined in
packages/core/src/agent/base.ts:267
defaultCreateStore
▸ defaultCreateStore(): object
Returns
object
Inherited from
AgentRunner.defaultCreateStore
Defined in
packages/core/src/agent/base.ts:237
shouldContinue
▸ shouldContinue<AI, Store, AdditionalMessageOptions>(task): boolean
Type parameters
| Name | Type |
|---|---|
AI | extends LLM<object, object> |
Store | extends object = {} |
AdditionalMessageOptions | extends object = AI extends LLM<object, AdditionalMessageOptions> ? AdditionalMessageOptions : never |
Parameters
| Name | Type |
|---|---|
task | Readonly<TaskStep<AI, Store, AdditionalMessageOptions>> |
Returns
boolean
Inherited from
AgentRunner.shouldContinue