No description
  • JavaScript 85.3%
  • Shell 14.7%
Find a file
2026-05-28 14:33:58 +01:00
.claude-plugin Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00
skills/mog Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00
src/hooks Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00
.gitignore Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00
install.sh Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00
README.md Initial commit: mogmode skill, hooks, and plugin manifest 2026-05-28 14:33:58 +01:00

mogmode

Highly compressed genz 10x'er coding voice for Claude Code. Slang and emoji as semantic compression, full technical accuracy preserved.

Personal fork of caveman — stripped of multi-mode infra, rebranded for one purpose.


Before / After

Before (default Claude Code):

Your component re-renders because you're creating a new object reference each render. Wrapping the inline object prop in useMemo will preserve the reference across renders and prevent the unnecessary re-render.

After (mog active):

inline obj prop = new ref each render 😭 wrap in useMemo 🩹


What it does

  • Drops articles, filler, pleasantries, hedging
  • Compresses prose words (DB/auth/config/req/res/fn/impl), uses arrows for causality (X → Y)
  • Slang/emoji replace english phrases when shorter (cooked vs "broken beyond easy fix", 🪲 vs "bug spotted")
  • State-marker emojis ( ⚠️ 🚀 📉 🪲 🛠) at line-start for fast scanning
  • Auto-clarity: drops to normal prose for security warnings, destructive ops, multi-step sequences
  • Preserves technical accuracy (code blocks, function names, error strings, API names: never abbreviated)

Install

git clone <repo-url> ~/Git/mogmode
cd ~/Git/mogmode
./install.sh

Restart Claude Code. On first session, Claude will offer to wire up the [MOGGING] statusline badge.


Usage

  • Auto-on every session (default). The SessionStart hook activates mog automatically.
  • /mog off — disable for current session
  • /mog — re-enable
  • Natural language: "start mogging" / "stop mogging" / "mog me" / "talk like a mog" / "normal mode"
  • Env var: MOG_DEFAULT_MODE=off — disable auto-activation entirely

What lives where

mogmode/
├── .claude-plugin/plugin.json   Claude Code plugin manifest
├── skills/mog/SKILL.md          The skill body (lexicon, rules, examples)
├── src/hooks/
│   ├── mog-activate.js          SessionStart hook (writes flag, injects ruleset)
│   ├── mog-mode-tracker.js      UserPromptSubmit hook (slash/NL triggers, per-turn reinforcement)
│   ├── mog-config.js            Shared module (safeWriteFlag, readFlag — security hardening)
│   └── mog-statusline.sh        Bash [MOGGING] badge
├── install.sh                   Symlink-based plugin install
└── README.md

Uninstall

rm ~/.claude/plugins/mogmode

Restart Claude Code. The mog plugin will no longer load.


Lineage

Hook security architecture (symlink-safe safeWriteFlag/readFlag, statusline whitelist + control-byte stripping) is inherited from caveman by Julius Brussee. Lexicon and mog rebrand are bespoke.