A lightweight, AI-friendly blog platform with semantic HTML, markdown support, and simple commenting.

• Semantic HTML structure optimized for AI crawling
• Markdown post authoring with preview
• Real-time comments with tripcodes
• Infinite scroll via HTMX
• SEO-optimized with sitemap, RSS feed, and JSON-LD
• Minimal, responsive design with Tailwind CSS
• SQLite database for simplicity
• Dream Window Integration - Live AI art streaming at /dreams

• Python 3.10+
• uv for dependency management

1. Clone the repository
2. Setup the development environment:

# Install dependencies
uv sync

Create a .env file with your configuration:

# Copy example configuration (if exists)
cp .env.example .env

# Edit with your settings
nano .env

Key environment variables:

• DATABASE_URL: SQLite or other database URL
• AETHERA_TRIPCODE_SALT: Salt for comment tripcodes
• DREAM_GEN_AUTH_TOKEN: Shared secret for GPU worker authentication (Dreams module)
• RUNPOD_API_KEY: RunPod API key for GPU lifecycle management (Dreams module)
• RUNPOD_ENDPOINT_ID: RunPod endpoint ID for the Dream Window worker (Dreams module)

The application uses SQLite with SQLModel and Alembic for migrations:

# Apply migrations to create the database
uv run alembic upgrade head

# Start the development server
uv run python -m aethera.main

The blog will be available at http://localhost:8000.

# Build the Docker image
docker build -t aethera:latest .

# Run the container
docker run -p 8000:8000 \
  -e AETHERA_TRIPCODE_SALT=your_secure_salt \
  -v ./data:/app/data \
  aethera:latest

• aethera/ - Main package
  • api/ - API routers for posts, comments, SEO, and dreams
  • models/ - SQLModel database models
  • templates/ - Jinja2 templates
  • static/ - Static files (CSS, JS, uploads)
  • utils/ - Utility functions
  • dreams/ - Dream Window streaming module

æthera hosts the web viewer for Dream Window, a continuously morphing AI art generator. Visit /dreams to watch live AI-generated art that never repeats.

The Dreams module acts as a relay between browsers and a GPU cloud worker:

Browsers ←→ æthera (VPS) ←→ RunPod GPU Worker
         WebSocket        WebSocket

• ViewerPresenceTracker: Manages GPU lifecycle based on viewer count
• FrameCache: Stores recent frames for instant display on connect
• DreamWebSocketHub: Broadcasts frames to all connected viewers
• RunPodManager: Starts/stops GPU workers on demand

Set these environment variables for Dreams functionality:

# Required for GPU worker authentication
DREAM_GEN_AUTH_TOKEN=your_secure_shared_secret

# Required for GPU lifecycle management
RUNPOD_API_KEY=your_runpod_api_key
RUNPOD_ENDPOINT_ID=your_endpoint_id

The GPU worker only runs when viewers are present:

1. First viewer connects → GPU starts (~30-60s cold start)
2. Frames stream to all viewers
3. Last viewer disconnects → 30s grace period → GPU stops
4. Pay only for actual usage (per-second billing on RunPod)

æthera includes a simple in-memory rate limiter to prevent comment spam. Important limitations:

• The rate limiter is in-memory and will reset when the server restarts
• In a multi-worker/process deployment, each worker maintains its own rate limit state
• For production with multiple workers or load balancing, consider implementing a Redis-backed rate limiter

Licensed under CC BY 4.0