agent-core / Exports / IChatOptions
Interface: IChatOptions
Options for AI provider chat requests
Hierarchy
↳
IChatOptions
Table of contents
Properties
Properties
openai
• Optional openai: Object
OpenAI specific options
Type declaration
| Name | Type |
|---|---|
organization? | string |
user? | string |
stop? | string | string[] |
presencePenalty? | number |
frequencyPenalty? | number |
logitBias? | Record<string, number> |
topP? | number |
n? | number |
stream? | boolean |
suffix? | string |
echo? | boolean |
bestOf? | number |
logprobs? | number |
Inherited from
IProviderSpecificOptions.openai
Defined in
packages/agent-core/src/interfaces/provider.ts:105
anthropic
• Optional anthropic: Object
Anthropic specific options
Type declaration
| Name | Type |
|---|---|
stopSequences? | string[] |
topP? | number |
topK? | number |
metadata? | { userId?: string } |
metadata.userId? | string |
Inherited from
IProviderSpecificOptions.anthropic
Defined in
packages/agent-core/src/interfaces/provider.ts:122
google
• Optional google: Object
Google specific options
Type declaration
| Name | Type |
|---|---|
candidateCount? | number |
stopSequences? | string[] |
safetySettings? | { category: string ; threshold: string }[] |
responseModalities? | ("TEXT" | "IMAGE")[] |
topP? | number |
topK? | number |
Inherited from
IProviderSpecificOptions.google
Defined in
packages/agent-core/src/interfaces/provider.ts:132
tools
• Optional tools: IToolSchema[]
Tool schemas to provide to the AI provider
Defined in
packages/agent-core/src/interfaces/provider.ts:156
maxTokens
• Optional maxTokens: number
Maximum number of tokens to generate
Defined in
packages/agent-core/src/interfaces/provider.ts:158
temperature
• Optional temperature: number
Temperature for response randomness (0-1)
Defined in
packages/agent-core/src/interfaces/provider.ts:160
model
• Optional model: string
Model to use for the request
Defined in
packages/agent-core/src/interfaces/provider.ts:162
onTextDelta
• Optional onTextDelta: TTextDeltaCallback
Callback for text deltas during streaming. When provided, the provider should use streaming internally and call this for each text chunk, while still returning the complete assembled message.