Interface ComposableVoiceAgentOptions

interface ComposableVoiceAgentOptions {
    llm: LanguageModel;
    stt: STTProvider;
    systemPrompt?: string;
    tts: string;
    ttsOptions?: SynthesizeOptions;
}

Properties

llm: LanguageModel

ai-sdk LanguageModel (e.g. openai("gpt-5.4-mini")).

systemPrompt?: string

Seeded as the first message in conversation history so the LLM has guidance before any user audio arrives. Defaults to ComposableVoiceAgent.DEFAULT_SYSTEM_PROMPT.

tts: string

TTS voice in "<provider>/<voiceId>" form.

Test seam — TTS options forwarded to synthesize.