Physical push-to-talk button for agentic coding: turned a cheap Bluetooth selfie remote into a clicker for Google Antigravity
A developer transformed a cheap Bluetooth selfie remote into a push-to-talk button for Google Antigravity, enhancing agentic coding. The remote, which registers as a Bluetooth HID keyboard emitting Volume Up/Down keystrokes, allows users to activate Google Antigravity's native speech-to-text dictation (Ctrl+M) remotely. This modification enables users to prompt from various locations, such as leaning back or pacing, rather than being confined to the keyboard, improving the convenience and purpose of voice prompting.
Unlike other voice prompting setups, this developer's solution repurposes a common Bluetooth selfie remote, offering a free and open method to enhance Google Antigravity's dictation.
时间与来源
时间显示为 UTC
显示时区：UTC
本地时区尚不可用，暂时显示 UTC。
发布当时偏移：UTC+02026年9月19日 09:53 UTC
收录当时偏移：UTC+02026年9月19日 14:00 UTC
- 发布
- 2026年9月19日 09:53
- 收录
- 2026年9月19日 14:00
- 来源类型
- 开发者社区
- 档位
- 社区
- 信源状态
- 正常
档位是按信源手工设定的编辑判断，不是逐条打分。
Google Antigravity has native speech-to-text dictation with Ctrl+M, and the transcription quality is surprisingly solid for voice prompting. But having to sit right at the keyboard just to hit Ctrl+M to start talking and then press Enter to submit defeated the entire purpose of voice prompting for me. I wanted to lean back, pace around the room, or sit on the couch while prompting.
I took an old generic Bluetooth selfie shutter remote (the standard $2 AB Shutter 3 button you find on Amazon/AliExpress). When paired to Windows over Bluetooth, these remotes register as standard Bluetooth HID keyboards that emit Volume Up or Volume Down keystrokes.
I wrote a lightweight Python daemon that runs silently in the background to handle the integration:
- Intercepts the remote keystroke at the Win32 driver level using a low-level keyboard hook (WH_KEYBOARD_LL).
- Suppresses the keypress so the Windows volume bar / HUD never flashes on your screen.
- Single click: Restores and focuses the Antigravity window across monitors or virtual desktops, and toggles the built-in microphone (Ctrl+M).
- Double click: Stops the mic if you are currently dictating, waits a fraction of a second for transcription to settle, and fires Enter (hardware scan code 0x1C) to submit the prompt to the agent.
- Shift + Click: Passes through the original volume keystroke so you can still adjust your PC volume normally when needed.
- Subtle audio beeps give confirmation through your speakers so you know whether you toggled the mic or submitted without looking at the monitor.
It uses around 15MB of RAM, runs silently via pythonw with no terminal window popping up, and has a one-click batch script to add it to Windows Startup so it works automatically whenever you boot up.
Repo with the code and setup guide: https://github.com/X7xenon/antigravity-bluetooth-remote
Free and open for anyone to use or tweak for their own workflow. Feedback is welcome.