WebSitter: Making the Web Safer for Everyone

JavaScript Chrome Extension (MV3) TensorFlow.js Groq
GitHub

WebSitter is a Chrome browser extension built to make the web safer and more accessible for the users it looks after least: children, elderly people, and anyone with low digital literacy. The idea came from watching a grandparent fall for online misinformation and virus-laden pop-ups, a reminder that the web offers almost no built-in safeguards for vulnerable users.

WebSitter is a toggleable extension with two dedicated modes, Elderly Mode and Kid-Safe Mode, each designed around the specific risks that group faces online.

Features

Elderly Mode

  • Enlarged Interface - Enlarges buttons and tap targets (44px) across pages to make navigation easier.
  • Ad & Scam Blocking - Blocks ad domains and surfaces warning banners when scam-like language is detected.
  • Tap-to-Explain - Lets users tap any button or link to get a plain-language explanation, powered by Groq.
  • Text-to-Speech - Reads selected content aloud using the browser's built-in Web Speech API.

Kid-Safe Mode

  • Image Filtering - Detects and blurs inappropriate images in real time using client-side NSFW detection, with click-to-reveal.
  • Unsafe Domain Warnings - Warns before a child navigates to an unvetted or unsafe website.
  • Toxic Content Filtering - Scores page text for toxicity and blurs flagged content automatically.

Technical Architecture

WebSitter is built on Chrome's Manifest V3 framework, with a background service worker handling API calls and declarative net request ad-blocking, and a content script injected into pages to drive on-page behavior such as enlarging elements, blurring images, and displaying warnings.


Image filtering runs entirely client-side using NSFW.js on top of TensorFlow.js, so no image data ever leaves the device. Plain-language explanations and toxicity scoring are powered by an LLM served through Groq, chosen for its low-latency inference to keep the tap-to-explain feature responsive.