Type Alias: UseAssistantProps
UseAssistantProps:
object
Defined in: packages/core/src/hooks/use-assistant.ts:16
Props for configuring the AI Assistant and useAssistant hook.
Type declaration
abortController?
optional
abortController:AbortController
Optional AbortController to cancel requests.
apiKey
apiKey:
string
The authentication key/token for the model provider's API. For example, how to get the OpenAI API key.
baseUrl?
optional
baseUrl:string
Optional base URL for API requests.
chatEndpoint?
optional
chatEndpoint:string
The server endpoint for handling chat requests (e.g. '/api/chat'). Required for server-side support.
description?
optional
description:string
Optional description of the assistant's purpose.
functions?
optional
functions:RegisterFunctionCallingProps
[] |Record
<string
,ExtendedTool
<any
>>
Deprecated
Use tools instead. Custom functions/tools the assistant can use, either as an array or record object.
historyMessages?
optional
historyMessages:Message
[]
The history of messages exchanged with the assistant.
instructions
instructions:
string
System instructions/prompt for the assistant.
maxSteps?
optional
maxSteps:number
Maximum number of steps/iterations in a conversation.
model
model:
string
The specific model identifier to use:
modelProvider
modelProvider:
string
The AI model provider:
- openai
- anthropic
- deepseek
- xai
- ollama
name
name:
string
The display name of the assistant.
temperature?
optional
temperature:number
Controls randomness in responses (0-1).
toolCallStreaming?
optional
toolCallStreaming:boolean
Whether to stream tool calls.
toolChoice?
optional
toolChoice:ToolChoice
<ToolSet
>
Controls how the assistant selects tools to use.
tools?
optional
tools:Record
<string
,ExtendedTool
<any
>>
Custom tools the assistant can use. E.g. { localQuery: localQueryTool }
topP?
optional
topP:number
Controls diversity of responses via nucleus sampling (0-1).
version?
optional
version:string
Optional API version to use.
voiceEndpoint?
optional
voiceEndpoint:string
The server endpoint for handling voice/audio requests.