Spoken output for Antigravity

Antigravity listens. Now it can talk back.

Antigravity includes live transcription for prompts. TalkToCursor adds the other half: concise spoken agent updates in a managed, ElevenLabs, or local voice.

Connect Antigravity View npm package
agent.output / speak
AntigravityVoice
COMPLETE THE LOOP

Keep Gemini transcription. Add the voice you want.

TalkToCursor does not replace Antigravity’s microphone. It gives the agent a speech tool it can call while it works.

01 / OUTPUT

Hear the agent

Get spoken starts, milestones, questions, errors, and completion summaries while you work elsewhere.

02 / CHOICE

Pick the speech stack

Use managed TalkToCursor Cloud, the ElevenLabs voice library, or private local Voicebox.

03 / SHARED

Use one configuration

The same TalkToCursor package and saved voice settings work across compatible MCP hosts.

ANTIGRAVITY SETUP

Add the existing MCP package.

Antigravity versions may store MCP configuration in different locations. Open the MCP Store and choose Manage MCP servers or View raw config to edit the active file.

Register TalkToCursor

Add this entry inside the active MCP configuration shown by Antigravity:

{
  "mcpServers": {
    "talktocursor": {
      "command": "npx",
      "args": ["-y", "talktocursor"]
    }
  }
}

Choose your speech provider

Open the shared settings app, then select TalkToCursor Cloud, ElevenLabs, or Voicebox.

npx -y --package talktocursor talktocursor-settings

Set voice behavior

Add these instructions as a Global Rule in ~/.gemini/GEMINI.md, or create an Always On Workspace Rule from Antigravity’s Customizations panel:

## Voice feedback

Use the `speak` tool at task start, at important
milestones, when asking a question, and at completion.
Keep spoken updates concise. Do not read code, paths,
logs, or stack traces aloud.

Restart and confirm

Fully quit and reopen Antigravity, confirm TalkToCursor is active in the MCP panel, then ask:

Use the speak tool to say:
"Antigravity voice is connected."
macOS PATH note: If Antigravity reports that npx cannot be found when launched from the Dock or Spotlight, replace "npx" with the absolute path returned by which npx. Running npx -y talktocursor once in Terminal can also warm the package cache.
WHAT STAYS NATIVE

A complement, not a replacement.

  • Use Antigravity’s mic for Gemini-powered live transcription and editable prompts.
  • Use TalkToCursor’s speak tool for automatic spoken task feedback.
  • Use your preferred provider when voice choice, local processing, or portable settings matter.
  • Use optional external dictation only if needed. Wispr Flow and Handy can provide system-level dictation, but the automatic TalkToCursor loop remains Cursor-focused.
DETAILS

Before you connect.

Does Antigravity already have voice?

Antigravity includes live transcription for voice input. TalkToCursor adds selectable text-to-speech output so the coding agent can speak updates aloud.

Do I need a separate TalkToAntigravity app?

No. Antigravity launches the same open-source TalkToCursor MCP package. Only the host configuration and instruction file differ.

Which MCP configuration path should I use?

Use the exact file opened by Antigravity’s MCP management screen. The path has varied between releases, so the in-app raw configuration is the reliable source.

Let Antigravity keep you in the loop.

Connect one MCP tool and choose how your agent sounds.

Install from npm Read the docs