Workout Gate

Workout Gate 🏋️
Your AI works hard, so should you.

A Claude Code and Codex hook that blocks your prompt until you work out —
push-ups or squats, counted live via webcam. When a challenge fires you pick
your pain (say 6 push-ups *or* 9 squats). Random reps, session-persistent debt
(no closing the tab to skip), streak stats, and three trigger modes.

*Version française : README.fr.md · made by @Botchet*

Requirements

- Python 3.9–3.13 — including the macOS system Python 3.9, so there's no newer Python to install.
- A webcam + an internet connection (first run downloads MediaPipe/OpenCV and a ~9 MB pose model).
- macOS for the zero-config plugin onboarding — it pops the setup in a Terminal and triggers the camera-permission dialog. Linux/Windows work too; Claude just points you at bootstrap.sh to run once by hand.
- git and python3 on your PATH.

Install

As a Claude Code plugin (CLI + desktop, recommended)

Then start a new session (or run /reload-plugins) — nothing happens
until you do. The plugin works in Claude Code CLI and desktop. Onboarding pops
up in a Terminal window on its own — dependencies install, then a 30-second
wizard (your max, trigger choice, a 2-pushup camera test). Until setup is done,
prompts pass freely. The gate and /workout-gate:workout then work in every
session, and plugin updates never break the install (the runtime lives in
~/.workout-gate/).

Also works in Codex (CLI + desktop)

The gate uses the same UserPromptSubmit hook in Codex — CLI and desktop.
From Codex, open /plugins, add the BotchetDig/workout-gate marketplace if
needed, then install workout-gate. Codex shows the hook review separately, so
approve the plugin hooks once with /hooks before testing a prompt.

Prefer terminal commands? After setup, wire it into all your Codex…