We got tired of losing our AI context, so my father and I built a zero-cloud visual data vault in pure JS

Every developer knows the pain of AI context loss. You have a massive, deep brainstorming session or debugging cycle inside ChatGPT, Claude, or Gemini. The second you refresh, open a new tab, or hit a token limit, that entire cognitive window vanishes.

My father (Chief Architect) and I wanted a solution that didn't rely on bloated cloud databases, paid subscriptions, or heavy server syncs.

So, we built Project Kaleidoscope (KVMS-1.0)—a universal, local-first Manifest V3 web extension framework that compresses, encrypts, and transforms deep chat text logs into a single, beautiful 512x512 image file. We call it a "Memory Glyph."

When you drop your custom Glyph back into a fresh, blank multimodal AI interface later, the extension aligns the matrix, prompts for a passkey PIN, and instantly rehydrates your session context completely offline.

🎨 Stacking Data in 3D: The 3-Axis Architecture

Instead of relying on fragile high-density pixel grids, our multiplexing layout engine crams hundreds of text kilobytes into an incredibly light visual footprint using three independent spatial dimensions simultaneously:

• Axis A (Shapes): We use shape parity rules. Is a drawn element a square or a circle? That geometric choice instantly tells the computer what type of system rule or schema it's parsing.
• Axis B (Coordinates): We map priorities. Exact X,Y pixel layouts establish chronological order and structural data dependencies.
• Axis C (Color Multipliers): By manipulating Red, Green, and Blue RGB channels, a single dot acts as a direct multiplier to encode background access variables.

To make sure you can scan or recover this from any digital screen or a physical piece of paper, the canvas features built-in Section 3 OpenCV Alignment Targets at its corner matrix pins.

🔒 100% Client-Side / Local-First

There is zero cloud. The AES-256-GCM hardware encryption and the canvas renderings execute entirely inside your local browser memory using native Web Crypto APIs. If the internet goes down, your data standard still works perfectly.

🚀 Open Source & Live on GitHub

We just pushed our core architecture preview live under our profile KVM-labs. Since this is an initial technical proof-of-concept, we would absolutely love your feedback on our pipeline math, browser timing hooks, and visual cryptography layout rules!

👉 Explore the Repository Here: https://github.com/KVM-Labs/Project-Kaleidoscope-KVMS-1.0#a-universal-high-density-visual-data-standard-for-offline-ai-context-restoration

Drop a comment below and let us know your thoughts on visual data vaulting!