0-H8

0-H8 is a Chrome extension that replaces toxic comments on social media with positive messages using an LLM. it's a filter for your peace of mind. Built to make your social media fun again.

Comment

Inspiration

Social media is where most people go for entertainment, but it's also where hate speech thrives. After seeing how negativity and toxic comments ruin the online experience for both users and content creators, I wanted to build a solution that doesn’t moderate or censor, but instead elevates the experience.

What it does

0-H8 is a Chrome extension that detects hate speech in Instagram comments and replaces it with heartful, positive messages using GPT-4. It works entirely on the client side, filtering toxicity without removing posts or reporting users, just making social media feel better, instantly.

How we built it

  • Built using JavaScript to scrape and process comments in the Instagram DOM
  • Sends comments via a POST request to a webhook
  • The webhook connects to GPT-4, which returns a classification (hateful or not)
  • Hateful comments are replaced in real-time with kind, uplifting messages
  • HTML and CSS are used for the popup and extension UI

Challenges we ran into

  • Identifying dynamic class names and comment containers on Instagram
  • Building a reliable and fast connection between the extension and the LLM API
  • Ensuring the UX remains smooth while processing multiple comments
  • Avoiding false positives without under-detecting subtle toxicity

Accomplishments that we're proud of

  • Successfully integrated GPT-4 into a real-time Chrome extension
  • Created a seamless user experience that improves the vibe of Instagram without altering the platform
  • Built a scalable architecture that can easily be applied to other social platforms

What we learned

  • Working with LLMs in real-world applications requires careful formatting and context awareness
  • Webhooks and automation tools can bridge the gap between browser-side logic and cloud-based AI
  • Even small UI changes can make a big difference in user perception and mental well-being

What's next for 0-H8

  • Expanding to support other platforms like YouTube, TikTok, and Reddit
  • Using more advanced models and providing context-aware analysis
  • Allowing users to customize the tone or style of the replacement messages
  • Adding light analytics to show users how much toxicity was filtered over time

Built With

Share this project:

Updates