YouTube Summarizer

A modular Gradio application that loads a YouTube transcript, summarizes it, and answers questions using retrieval-augmented generation.

Architecture

• youtube_app/config.py loads and validates environment settings.
• youtube_app/transcript.py parses YouTube URLs and normalizes captions.
• youtube_app/providers.py contains OpenRouter LLM and embedding adapters.
• youtube_app/rag.py chunks transcripts and performs direct FAISS retrieval.
• youtube_app/chains.py owns prompts and LCEL chains.
• youtube_app/service.py coordinates the application use cases and caches indexes per video.
• youtube_app/ui.py contains only the Gradio presentation layer.
• main.py is the deployment entry point.

Local setup

Use Python 3.11 or newer. From this directory:

python -m venv .venv
source .venv/bin/activate
python -m pip install -e ".[dev]"
cp .env.example .env

Set OPENROUTER_API_KEY in .env, then run:

python main.py

Open http://127.0.0.1:7860 in a browser. The free router is the default chat model; set OPENROUTER_MODEL to another OpenRouter model when needed.

Tests

pytest

The tests use fake embeddings and do not call YouTube or OpenRouter.

Docker

docker build -t youtube-summarizer .
docker run --rm -p 7860:7860 --env-file .env youtube-summarizer

GitHub safety

.env is ignored and must never be committed. If an API key has ever been shared or committed, revoke it and create a replacement before publishing the repository.