Getting Started

Prerequisites

• Conda or Miniconda
• Node.js 22+
• (Optional) CUDA GPU for accelerated OCR/embeddings

Quick Setup

git clone git@github.com:sylvanding/omelette.git && cd omelette

# Backend
conda env create -f environment.yml && conda activate omelette
cp .env.example .env
cd backend && alembic upgrade head
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000 &

# Frontend
cd ../frontend && npm install && npm run dev -- --port 3000

Open http://localhost:3000.

Configuration

Key environment variables in .env:

Variable	Purpose	Default
LLM_PROVIDER	LLM backend	mock (no API key needed)
OPENAI_API_KEY	OpenAI key	(empty)
ANTHROPIC_API_KEY	Anthropic key	(empty)
DATABASE_URL	Database path	sqlite:///./data/omelette.db
DATA_DIR	File storage	data
PDF_PARSER	PDF engine	auto

With LLM_PROVIDER=mock, no API keys are required for development.

First Steps

1. Create a project → Knowledge Bases → New
2. Add papers → Search & Add or upload PDFs
3. Explore → Browse, search, analytics
4. Chat → Ask questions about your literature in Playground

Project Structure

omelette/
├── backend/     # FastAPI (Python 3.12) · 861 tests
├── frontend/    # React SPA (TypeScript) · 273 tests
├── e2e/         # Playwright · 39 tests
├── docs/        # VitePress documentation
└── scripts/     # Ralph agent workflow