[Feature Request] Native audio input support in Claude Code VS Code / Cursor extension

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Feature Request: Voice / Audio Input in Claude Code Extension

Proposed Solution

Summary

Add native microphone / audio recording support to the Claude Code extension for VS Code, Cursor, and compatible forks, so users can dictate prompts instead of typing them.

Problem

Currently, the Claude Code extension only accepts text input. Competitors have already shipped voice input:

Cursor — built-in voice dictation
ChatGPT / Codex (OpenAI) — voice input in their tools
Gemini (Google) — audio input support
Claude Code has no equivalent, which is a noticeable gap for users who prefer or rely on voice.

Proposed Solution

Add a microphone button to the Claude Code extension chat input that:

Records audio from the user's microphone
Transcribes it (via Whisper or Claude's own audio capabilities)
Inserts the transcribed text into the prompt field (or sends directly)
Optionally: support sending raw audio to Claude's multimodal API directly.

Why This Matters

Speeds up interaction — speaking is faster than typing for many users
Accessibility — essential for users with motor disabilities
Mobile/tablet use cases
Natural fit given Claude already supports audio in the API
Environment

VS Code / Cursor / Windsurf and other VS Code forks
macOS / Windows / Linux
Additional Context

This would make Claude Code significantly more competitive with other AI coding assistants and bring the extension in line with Claude's own multimodal capabilities.

Alternative Solutions

_No response_

Priority

High - Significant impact on productivity

Feature Category

Developer tools/SDK

Use Case Example

_No response_

Additional Context

Additional Context

This would make Claude Code significantly more competitive with other AI coding assistants and bring the extension in line with Claude's own multimodal capabilities.

To lower the implementation barrier, the extension could leverage OpenAI's Whisper — a proven, open-source speech recognition model that runs locally or via API:

GitHub: https://github.com/openai/whisper
Supports 99 languages, works offline, MIT licensed
Already widely used in VS Code extensions for transcription
Using Whisper would mean no dependency on a third-party transcription service — audio stays local, which also addresses privacy concerns that developers care about.

Alternatively, if Anthropic adds native audio input to the Claude API (which the API already partially supports), the extension could pipe audio directly to Claude without any third-party dependency at all.

Either path is technically straightforward — the hard part is the UX (microphone button, recording state, error handling), not the transcription itself.