Are You Outsourcing Your Thinking?
Test your prompt below to see if you're using AI productively or bypassing thinking you could easily do yourself.
🔍 Prompt Analyzer Sandbox
📊 How Prompts Are Classified
This web demo runs a lightweight, on-device AI model in your browser (compiled via WebAssembly) that scores prompt intent into three distinct categories:
Deep research, code architecture, technical debugging, synthesis, and learning concepts. You are driving the work and using AI as a tool.
Quick arithmetic, routine formulation, or questions you could easily recall or compute with minimal effort.
Riddles, social decision-making ("what should I say?"), or asking AI to problem-solve before you've attempted it yourself.
🛡️ What is the MindShield Extension?
This sandbox runs the same classification engine used in MindShield, a lightweight browser extension for ChatGPT, Claude, Google Gemini, and Grok.
Instead of blocking AI, the extension automatically intercepts mental shortcuts in real time and inserts a brief reflection pause: asking "Do you actually need AI for this?" before you submit. Everything runs 100% locally on your machine with zero server telemetry.