Setup Instructions

1. Create a .env file. Instructions on how to get the CLIENT_ID and ACCESS_TOKEN below.

   TWITCH_CLIENT_ID=CLIENT_ID
   TWITCH_ACCESS_TOKEN=ACCESS_TOKEN
   TWITCH_CHANNEL_USERNAME=YOUR_USERNAME
   

2. Generate Access Token:
   • Go to this website: https://twitchtokengenerator.com/
   • Scroll down to Generated Tokens
   • Click Bot Chat Token
   • Click Authorize
   • Solve Captcha
   • Copy the two variables: TWITCH_CLIENT_ID and TWITCH_ACCESS_TOKEN to the .env file

Examples

Twitch Chat Listening

python3 -m src.examples.twitch_chat_listener

Ollama Chat Testing

python3 -m src.examples.ollama_chat_example

YouTube Transcript with Segments

python3 -m src.examples.youtube_segments_example

YouTube Video Chunk Scoring

python3 -m src.examples.score_video_chunks_example

Download Top YouTube Clips (Scored & Downloaded)

python3 -m src.examples.download_top_clips_example

Project Structure

clint/
├── src/
│   ├── services/          # API clients (Twitch, YouTube, LLM)
│   │   ├── llm/           # LLM service and prompts
│   │   ├── transcript_processor.py # Transcript extraction and segmentation
│   │   ├── twitch_client.py
│   │   └── youtube_client.py
│   └── examples/          # Example scripts

Running Model Locally

1. Install ollama follow these instructions
2. Start the server by running: ollama serve
3. Pull the LLM model you'll want to test with. Example: ollama pull llama3.1:8b
4. Run the ollama chat example. Instructions can be found under Ollama Chat Testing

Troubleshooting

If you get the following error: "Requested format is not available. Use --list-formats for a list of available formats" try updating the yt-dlp package by running: python -m pip install -U yt-dlp