A pixel-art AI companion that lives on your macOS Dock. Rocky paces back and forth at the base of your screen and responds to questions via a local LLM. When you open a distracting website, Rocky yells at you until you leave.

• macOS (required)
• Node.js 18 or later
• Ollama for local LLM inference

1. Clone and install dependencies

git clone https://github.com/akarvir/little-rocky.git
cd little-rocky
npm install

2. Install and start Ollama

Download Ollama from ollama.com/download/mac and run it. Then pull a model:

ollama pull qwen2.5:7b

Any model listed by ollama list will work. qwen2.5:7b is a good balance of speed and quality on Apple Silicon.

3. Configure environment

cp .env.example .env

The defaults use Ollama with llama3. Update VITE_LLM_MODEL to match whichever model you pulled:

VITE_LLM_PROVIDER=ollama
VITE_LLM_MODEL=qwen2.5:7b

To use Claude instead, set VITE_LLM_PROVIDER=claude and add your VITE_ANTHROPIC_API_KEY.

4. Run

npm run dev

Rocky will appear just above the Dock and begin pacing.

Rocky yells when you visit Twitter, YouTube, Instagram, TikTok, or X. To enable this:

1. Open Chrome and go to chrome://extensions
2. Enable Developer mode (toggle in the top right)
3. Click "Load unpacked" and select the extension/ folder in this repo

The extension connects to Rocky automatically when the app is running and reconnects if Rocky restarts.

Edit .env at any time and restart npm run dev:

# Local (default)
VITE_LLM_PROVIDER=ollama
VITE_LLM_MODEL=qwen2.5:7b

# Claude API
VITE_LLM_PROVIDER=claude
VITE_LLM_MODEL=claude-haiku-4-5-20251001
VITE_ANTHROPIC_API_KEY=sk-ant-...