Sharpie

Self-hostable AI prompt playground with local LLM support

Build, test, and share AI prompts with zero API costs. Run entirely on your machine with Docker.

Features

• One-command setup - docker-compose up and you're running
• Fully self-hosted - Your prompts never leave your machine
• Zero API costs - Uses local Ollama models (qwen2.5:3b by default)
• Share & Fork - Generate shareable URLs for any prompt
• Real-time streaming - Watch AI responses generate live
• Markdown rendering - Beautiful formatted responses with syntax highlighting
• GPU accelerated - Leverages your NVIDIA GPU automatically
• No dependencies - Everything runs in Docker containers

Screenshots

Main Interface

Response

Fork & Share

Backend

Settings

Quick Start

Prerequisites

• Docker Desktop installed
• 10GB free disk space (for Ollama model)
• (Optional) NVIDIA GPU with CUDA support

Installation

# Clone the repository
git clone https://github.com/heyrtl/sharpie.git
cd sharpie

# Start all services
docker-compose up --build

That's it! Open http://localhost:5173 in your browser.

First run takes 5-10 minutes to download the Qwen2.5-3B model (~2GB).

Usage

Basic Workflow

1. Write your prompts - System and user prompts in the editor
2. Run - Press Cmd/Ctrl + Enter or click "Run Prompt"
3. Share - Click "Share" to get a shareable URL
4. Fork - Click "Fork" to create a copy and modify

Sharing Prompts

Share URLs like http://localhost:5173?p=abc123 with anyone running Sharpie. They can:

• View your prompt
• Run it with their local model
• Fork and modify it

Switching Models

1. Click the settings icon
2. Select from available Ollama models
3. Models are auto-detected from your Ollama instance

Architecture

Frontend (React + Vite)
    ↓
Backend (FastAPI)
    ↓
Ollama (Local LLM)
    ↓
SQLite (Prompt Storage)

• Frontend: React app with real-time streaming UI
• Backend: FastAPI server handling prompts and streaming
• Ollama: Local LLM inference with GPU acceleration
• SQLite: Embedded database for saved prompts

Configuration

Using Different Models

Pull any Ollama model:

docker exec -it sharpie-ollama ollama pull llama3.2:3b
docker exec -it sharpie-ollama ollama pull mistral:7b

Then select it in Settings.

Environment Variables

Copy .env.example to .env and customize:

OLLAMA_HOST=http://ollama:11434
DATABASE_PATH=/app/data/sharpie.db

GPU Support

GPU is auto-detected. To disable GPU and run CPU-only, remove the deploy section from docker-compose.yml.

Development

Project Structure

sharpie/
├── backend/           # FastAPI server
│   ├── main.py       # API routes
│   ├── database.py   # SQLite handlers
│   ├── models.py     # Pydantic models
│   └── utils.py      # Helpers
├── frontend/          # React app
│   └── src/
│       ├── App.jsx
│       ├── components/
│       └── utils/
└── docker-compose.yml

Running Locally (without Docker)

Backend:

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

Frontend:

cd frontend
npm install
npm run dev

Make sure Ollama is running separately.

Troubleshooting

Port Already in Use

Change ports in docker-compose.yml:

ports:
  - "8001:8000"  # Backend
  - "5174:5173"  # Frontend

Ollama Model Not Downloading

Manually pull the model:

docker exec -it sharpie-ollama ollama pull qwen2.5:3b

GPU Not Detected

Check NVIDIA Docker runtime:

docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi

If it fails, you may need to install nvidia-container-toolkit.

Out of Disk Space

The Ollama model requires ~2GB. Free up space or use a smaller model:

docker exec -it sharpie-ollama ollama pull qwen2.5:0.5b

Contributing

Contributions are welcome! Here's how:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Development Setup

See CONTRIBUTING.md for detailed setup instructions.

Roadmap

• Multi-model API support (OpenAI, Claude, Gemini)
• Prompt versioning and history
• Collaborative editing
• Export prompts as JSON
• Prompt analytics
• Browser extension
• Model comparison view

Security

See SECURITY.md for security considerations and best practices.

License

MIT License - see LICENSE for details.

Author

Ratul Rahman (@heyrtl)

• Website: ratul-rahman.com
• GitHub: @heyrtl
• Twitter: @heyrtl

Acknowledgments

• Ollama for local LLM inference
• FastAPI for the backend framework
• React for the frontend
• Qwen team for the excellent small language models

Star History

If you find this useful, consider giving it a star!

---

Built with care for the prompt engineering community.