RLM Playground

A fullstack playground for the Recursive Language Model using Next.js frontend, FastAPI backend and DsPy.

Quick Start (Docker)

# Copy environment file
cp apps/frontend/.env.example apps/frontend/.env
cp apps/backend/.env.example apps/backend/.env

# Edit .env with your API keys
# Then build and run
docker compose up --build

• Frontend: http://localhost:3000
• Backend API: http://localhost:8000

Development

Requirements: Node.js 20+, Python 3.12+, uv

# Install frontend dependencies
npm install

# Install backend dependencies
cd apps/backend && uv sync

# Start all services
npm run dev

Environment Variables

Set in apps/backend/.env:

Variable	Description	Example
DSPY_DEFAULT_MODEL	LiteLLM model identifier	openai/gpt-4o
OPENAI_API_KEY	OpenAI API key	sk-...
OPENAI_API_BASE	Custom endpoint (optional)	http://localhost:8080/v1
OPENROUTER_API_KEY	OpenRouter key (if using)	sk-or-...
ANTHROPIC_API_KEY	Anthropic key (if using)	sk-ant-...

Project Structure

apps/
├── frontend/    # Next.js 16, React 19, Tailwind
└── backend/     # FastAPI, DSPy, uv

Docker Images

Service	Build	Runtime
Backend	ghcr.io/astral-sh/uv:python3.11	gcr.io/distroless/python3-debian12:nonroot
Frontend	ubi10/nodejs-24-minimal	registry.access.redhat.com/ubi10/nginx-126