hugo-alves

TokenGochi — Development skill for Claude Code

Development community

A mildly dystopian ESP32 virtual pet fed by your Claude Code and Codex CLI token usage.

How to install TokenGochi

This entry records only its repository, not the path inside it, so there is no exact command to give. Open hugo-alves/TokenGochi and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What TokenGochi does

A mildly dystopian ESP32 virtual pet fed by your Claude Code and Codex CLI token usage.

Alternatives in Development

  • Ccusage — CLI for analyzing Claude Code/Codex usage from local JSONL files 11.8k ★
  • Claude Usage — A local dashboard for tracking your Claude Code token usage, costs, and session history 864 ★
  • Caveman Claude Skill — Ultra-compressed communication mode for Claude 396 ★

README

Token Tamagotchi 🐣

**Give your AI token meter a face, a mood, and the power to make you feel slightly guilty.**

![AI coding tokens feeding the firmware-rendered TokenGochi pet on an ESP32 watch](docs/assets/readme/tokengochi-hero.png)

TokenGochi turns Claude Code and Codex CLI usage into a virtual pet living on a small ESP32 device. Your coding sessions feed it. As your token usage changes, the creature gets hungry, happy, sleepy, or sick—and quietly judges you from your desk or wrist.

Most token tracking lives in logs and account dashboards: useful numbers, but easy to forget until a limit suddenly matters. TokenGochi makes that invisible activity physical, ambient, and glanceable. You can see the shape of your AI coding habit without opening another tab.

The joke is a mildly dystopian reversal of the Tamagotchi: instead of keeping a digital creature alive by pressing buttons, you keep it alive by asking an AI to write code. Productive workflow or emotional blackmail from a tiny computer? That is between you and your pet.

![The four current firmware pet moods: happy, hungry, sleepy, and sick](docs/assets/readme/tokengochi-moods.png)

The project has three main pieces:

  • firmware that draws and animates the pet on an M5Stack StopWatch Dev Kit or a compact M5StickC Plus2-class device;
  • a local Node bridge that counts Claude Code and Codex CLI tokens and turns activity into pet state;
  • an optional Cloudflare backend for HTTPS access, plus optional Groq transcription so you can talk to the pet.
         ⌚  ← the StopWatch
   🍔  ⬆ tokens  ──▶  🌕 happy / 😟 hungry / 💤 sleepy / 🤒 sick
   ⬇ text         ⬆  466x466 round AMOLED, 4 mood-sprite frames + blink
  🌉 bridge/worker⌨  B voice mode → B record/send → Whisper on Groq → text
  ~/TokenGochi    📡  HTTPS API, fallback LAN mode available

Hardware

  • M5Stack StopWatch Dev Kit (C152, ESP32-S3R8, 8 MB PSRAM, 16 MB flash) — has 1.75" round AMOLED, 2 buttons, MEMS mic, speaker,