💡 Inspiration
Live coding streams often face context breaks when streamers encounter runtime bugs or syntax errors. Viewers frequently suggest fixes across chaotic chat messages. Twitch AI CodeHelper bridges this gap by letting streamers and viewers diagnose, debug, and explain code logic on the fly without leaving the live stream.
✨ What it does
- Real-Time Bug Resolution (!fixbug): Audience and streamers can submit runtime errors or broken snippets directly in chat to get concise solutions.
- Strict Twitch-Format Output: AI responses are optimized under 250 characters to prevent chat flooding while delivering actionable code fixes.
- Instant Event-Driven Chat Handling: Built with lightweight event loops handling Twitch IRC directly.
🛠️ How we built it
- Runtime: Node.js
- Twitch Integration:
tmi.js(Twitch IRC client) - AI Engine: Generative AI API (Low-latency streaming model)
- Environment: Dotenv for secret management
🧗 Challenges we ran into
Handling real-time chat rate limits and ensuring that AI responses stay concise enough for Twitch's character limit without breaking code syntax blocks.
🏆 Accomplishments that we're proud of
Successfully connecting a live Twitch IRC client event listener to an AI model and getting instant, accurate code fixes directly inside the stream chat room.
📚 What we learned
Gained deep insights into event-driven programming with Node.js, real-time socket/IRC communication, and prompt engineering for low-latency coding assistants.
Log in or sign up for Devpost to join the conversation.