Skip the bakwas ("nonsense") — extract and summarize YouTube video transcripts with the LLM of your choice.
Bakwas is a small, self-hosted Flask app that pulls captions from a YouTube video, sends them through the LLM provider you configure (Anthropic, OpenAI, Google, DeepSeek, Groq, OpenRouter, a local Ollama, or any OpenAI-compatible endpoint), and stores the summary in a local database so you can revisit it later.
- Works with any LLM via a pluggable provider registry — OpenAI-compatible endpoints are fully supported
- Concise (bulleted) or comprehensive (paragraph) summary styles
- Server-side sorting, pagination, and URL-based caching so you don't pay to regenerate the same summary twice
- Settings for default model, default summary style, and cost usage
- Ships as a single Docker container with a SQLite volume
git clone https://github.com/alifbae/bakwas.git
cd bakwas
cp .env.example .env
# Set at least ANTHROPIC_API_KEY (or another provider key)
docker-compose up -dBakwas is now available at http://localhost:5000.
Comprehensive docs are at docs.bakwas.alifbae.dev, including:
- Getting started with Docker
- Local development setup
- Environment variables
- Provider configuration
- Rate limiting
- Reverse proxy setup
- Troubleshooting
MIT — see LICENSE.