Skip to content

arach/speakeasy

v0.1.0MIT

Give coding agents a controllable voice through the native SpeakEasy menu-bar player.

GitHubOpen Website

SpeakEasy plugin

SpeakEasy gives Codex and ChatGPT Work a controllable voice through a permanent native macOS menu-bar player. It bundles the SpeakEasy CLI runtime and does not require a global SpeakEasy npm installation.

What it does

  • Generates speech with the zero-configuration macOS system voice or a configured cloud provider.
  • Autoplays or queues narration in the signed SpeakEasy menu-bar app.
  • Controls pause, resume, stop, skip, scrubbing, volume, and playback speed.
  • Shows a word-synced HUD and can return to the originating Codex task when a task identifier is available.

Requirements

  • macOS 14 or newer
  • Bun 1.0 or newer
  • The signed SpeakEasy app, which the bundled runtime can download from the official GitHub release when the user asks to install or use SpeakEasy

System speech needs no account or API key. Optional OpenAI, ElevenLabs, Groq, and Gemini voices use credentials configured locally by the user.

Privacy

Narration uses the local macOS voice by default. When a cloud provider is explicitly configured or selected, the requested text is sent directly to that provider. See the privacy policy and terms.

Support

Use the documentation or file an issue in the SpeakEasy repository. Do not include API keys in an issue.

Build the submission bundle

From the repository root, run bun run package:plugin-submission. The command rebuilds the bundled runtime, verifies the skill in an isolated extracted directory, and writes the versioned skill ZIP and its SHA-256 file under dist/. Complete RELEASE-CHECKLIST.md before uploading that ZIP.