kandor

A Chrome extension that protects kids from online predators, groomers, scammers, and manipulators — in real time.

What it does

Kandor runs quietly in the background as you browse supported social platforms. It scans your conversations for signs of manipulation, grooming, and other online threats. When it detects something suspicious, it surfaces a clear alert with:

• Risk & confidence scores — how dangerous and how certain
• Flagged messages — exactly what was flagged and why
• Threat categories — e.g. Trust grooming, Personal info request, Move off-platform
• Recommended action — e.g. Block and alert guardian

Why we built it

Online grooming and manipulation happen gradually and are hard to spot in the moment — especially for younger users. Kandor acts as a real-time safety layer on the platforms students already use, catching patterns that are easy to miss and giving parents and schools the information they need to act.

Vision:

We're exploring kandor as a tool for K-12 and school-managed devices — giving schools and parents a lightweight, real-time safety layer on the platforms students already use.

Tech Stack

Frontend / Extension

• Chrome Extension APIs (Chromium / Edge compatible)
• JavaScript, HTML, CSS

Backend

• Python, FastAPI

AI / NLP

• Rule-based NLP for fast, lightweight pattern detection
• Gemma 4 for deeper language understanding
• K2 Think V2 (70B reasoning model by IFM) as the core reasoning engine — used for multi-step threat analysis across complex, ambiguous conversations

Originally Built at

Hacktech 2026 — California Institute of Technology

Contributors: Chris Rios, Mario Chong, Mira Bhakta, Jericho De Guzman

Links

• 🌐 Landing page: kandor.tech
• 📋 Devpost: devpost.com/software/kandor