Spokenly
Spokenly for developers

Voice input for your coding agent

Press a shortcut, describe the change, and the text appears at your cursor. Spokenly works in Claude Code, Codex, Cursor, your terminal, and every other app on Mac, Windows, and Linux.

Download Spokenly Read the developer guide 
$ brew install spokenly

Homebrew on macOS. Windows and Linux installers are on the download page.

Works where you write code

Claude Code

Speak long prompts into the CLI. On macOS, Claude Code can ask you a question through MCP and continue when you answer by voice.

Claude Code setup

Codex

Describe the task out loud in the Codex CLI. On macOS, Codex connects to Spokenly over HTTP MCP for voice input during a session.

Codex setup

Cursor

Dictate into chat, Composer, and Cmd+K or Ctrl+K on macOS, Windows, and Linux.

Cursor setup

Terminal and editors

Text goes to the cursor, so the same shortcut works in any terminal, VS Code, or browser tab.

Documentation

Built for technical speech

Pick the model

Run Whisper or Parakeet on your device, or use cloud models such as GPT-4o Transcribe and Deepgram Nova.

Teach it your vocabulary

The dictionary helps supported models recognize names and technical terms. Word replacements fix the rest, for example claud to Claude.

Keep audio on your machine

With local models, audio stays on your device. Local Only Mode blocks network activity except local model downloads.

Let the agent ask you

The macOS sideload build runs a local MCP server. The agent calls it when it needs your input, you answer by voice, and the text goes back automatically.

Start in a minute

  1. Install Spokenly. Run brew install spokenly on macOS, or get the installer for Mac, Windows, or Linux from the download page.
  2. Put the cursor in the prompt. Any agent, terminal, or editor works.
  3. Press the shortcut and speak. Right Command on macOS, Ctrl+Space on Windows and Linux.

Questions developers ask

Is Spokenly free?

Yes. Local models are free with no account and no time limits, and cloud transcription with your own API keys costs nothing extra. Pro is an optional subscription that adds managed cloud models. See pricing.

Is it a transcription API?

No. Spokenly is a desktop dictation app. It types what you say into the app you are using.

Is it open source?

No. Spokenly is proprietary and has no public GitHub repository. Read more about local models and privacy.

Which platforms does it support?

macOS, Windows, Linux, iPhone, and a Chrome extension. The MCP server is available in the macOS sideload build only.