An Open Source AI Agent for YouTube Streamers
I've been experimenting with building small tools that make creators' lives easier, and one thing that stood out: YouTube streamers often struggle to keep up with live chat. So I built something to help: an AI agent that continuously monitors comments and responds intelligently on their behalf. It’s called AI YouTube Chat, and it’s fully open source. Repo: https://github.com/arberrexhepi/ai_youtube_chat What it does This isn’t just a chatbot…it’s an agent. It runs in the background during streams, watching the chat in real time, and sending context-aware replies using an LLM. That way, streamers can focus on the content without missing out on engaging with their viewers. The app is built in React and currently supports both OpenAI and Ollama endpoints. You can easily run it with local models or plug it into your OpenAI API key. Customizing the Agent The cool part? You can tweak the system prompt (aka the system role) in the LLM API call. This means anyone who clones the repo can customize how the agent behaves. Whether you want it to sound like you, act as a helpful moderator, or even provide extra context about your channel or stream, the choice is yours to make. Just drop in your own prompt, and the agent will start thinking in that voice. It’s one of the best ways to make the tool feel like an actual extension of your persona or brand. Feel free to clone it, use as is or extend the functionality of the agent’s app… or monetize it, it’s completely open source.

I've been experimenting with building small tools that make creators' lives easier, and one thing that stood out: YouTube streamers often struggle to keep up with live chat. So I built something to help: an AI agent that continuously monitors comments and responds intelligently on their behalf.
It’s called AI YouTube Chat, and it’s fully open source.
Repo:
https://github.com/arberrexhepi/ai_youtube_chat
What it does
This isn’t just a chatbot…it’s an agent. It runs in the background during streams, watching the chat in real time, and sending context-aware replies using an LLM. That way, streamers can focus on the content without missing out on engaging with their viewers.
The app is built in React and currently supports both OpenAI and Ollama endpoints. You can easily run it with local models or plug it into your OpenAI API key.
Customizing the Agent
The cool part? You can tweak the system prompt (aka the system role) in the LLM API call. This means anyone who clones the repo can customize how the agent behaves. Whether you want it to sound like you, act as a helpful moderator, or even provide extra context about your channel or stream, the choice is yours to make.
Just drop in your own prompt, and the agent will start thinking in that voice. It’s one of the best ways to make the tool feel like an actual extension of your persona or brand.
Feel free to clone it, use as is or extend the functionality of the agent’s app… or monetize it, it’s completely open source.