[Cursor SDK] Native audio attachments (parity with SDKImage)

Ideas Feature Requests
1

Feature request for product/service

Cursor SDK

Describe the request

Feature request for product/service

Cursor SDK

Describe the request

Please add native audio attachments to the Cursor SDK (@cursor/sdk / cursor-sdk), with parity to the existing image path (SDKImage / UserMessage.images).

Today, UserMessage supports text + images only. The IDE has Voice Input, but that is client-side speech-to-text before send — it is not exposed as an SDK/API attachment. Passing an audio file path in the prompt also does not work as multimodal input: the agent can only treat it as binary via tools, not as audio the model can hear.

Proposed API shape (illustrative)

TypeScript:

const run = await agent.send({
  text: "Transcribe and assess this reading",
  images: [/* existing */],
  audio: [{ data: base64Wav, mimeType: "audio/wav" }],
  // or: audio: [SDKAudio.fromFile("reading.wav")]
});


PYTHON:
run = agent.send(
    UserMessage(
        text="Transcribe and assess this reading",
        audio=[SDKAudio.from_file("reading.wav")],
    )
)




### Operating System (if it applies)
Linux
5

Hey @Lade!

Thanks for the suggestion. I’m not sure that any of the models you can currently use with Cloud Agents support audio as an input. Until such time as they do, I don’t think adding audio attachments would provide much value. Is there something I’m missing?