This application generates chess quizzes based on YouTube chess video transcripts. It uses AI to analyze the transcript and create relevant questions with chess positions.
- Fetch YouTube video transcripts
- Generate chess quizzes based on video content
- Display chess positions using FEN notation
- Interactive quiz interface
This is the main server file. It sets up an Express.js server and handles the following:
- Serves the static files for the frontend
- Provides API endpoints for fetching YouTube transcripts and processing them
- Interfaces with the 79AA Anthropic API to generate quiz questions
This file contains the frontend logic for the quiz interface. It handles:
- Displaying quiz questions and chess positions
- Managing user interactions and scoring
- Transitioning between questions
This file contains the main frontend JavaScript code. It's responsible for:
- Handling user input for YouTube video URLs
- Fetching transcripts and initiating quiz generation
- Managing the overall flow of the application on the client side
This file sets up a proxy server to handle CORS issues when fetching YouTube transcripts. It:
- Forwards requests to YouTube's transcript API
- Handles any necessary request modifications or error handling
This file provides a service layer for interacting with the Anthropic Claude API. It:
- Manages API requests to Claude
- Handles response parsing and error handling
- Provides a clean interface for other parts of the application to use Claude's capabilities
- Clone the repository
- Install dependencies with
npm install - Set up environment variables (API keys in .env)
- Run npm start (uses start_servers.mjs to run app.mjs and proxy_server.mjs)
- Enter a YouTube chess video URL
- Click "Generate Quiz"
- Answer the generated chess questions
- View your score at the end of the quiz
- Express.js
- Anthropic API
- Chess.js (for chess logic)
- Chessboard.js (for chess board visualization)
- Main issues is that FEN and quiz questions don't always match
- Quiz answers aren't checked. Need to use a LLM API call to check if users answer is close enough
This project is licensed under the MIT License.