I just realized when writing the previous post that my HomeLAB was originally created from seeing and using Gemini CLI with MD (MarkDown) files -- these files allowed the AI to retain memory without the cognitive drift (hallucinations).

So here is another update to the HomeLAB -- no more repeating yourself -- just type your query once and everything is placed on the Blackboard -- when you click on a model the first thing it will do is look at the Blackboard.md file for current data and continue with your request.
The Blackboard.md file is stored in your Vault-AI folder.

Also, as long as you do not clear the Blackboard (which would start a fresh version -- cleanslate), you can shut down the HomeLAB and everything is still on the Blackboard when you start your HomeLAB up again -- game-changer.
Best
Xard777

### New in v3.3: Neural Bridge (Shared Brain)
- **Shared Blackboard Protocol**: All departments (Strategy, Review, Local) now read and write to a single source of truth (`BLACKBOARD.md`).
- **Zero Copy-Paste**: Context is automatically shared. Ask Strategy to plan, then simply tell Local to "Read the Blackboard and proceed."
- **Project Lifecycle Management**: Use the new ` Clear Blackboard` button to archive your current project to a ZIP file and reset for a fresh start.

### Installation Steps
1. **Create a Folder**: Create a folder on your computer (D: or C: drive) called `VAULT-AI`.
2. **Download & Unzip**: Download this HomeLAB Zip and unzip it into `VAULT-AI`.
3. **Launch Antigravity**: "Add Folder to Workspace" ➔ Select `VAULT-AI`.
4. **Automated Setup**: Ask Antigravity: "Set up my HomeLAB".

### The Quad Architecture (Parallelized)
HomeLAB v3.3 uses a resilient "N+1" redundancy system:
- **Strategy Dept**: `Gemini 3 Flash` (Primary) ➔ `Gemini 2.5 Flash` (Fallback).
- **Antigravity Dept**: `Gemini 3.1 Pro` (Primary) ➔ `Gemini 3 Pro` (Fallback).
- **Review Dept**: `Kimi k2.5` (Primary) ➔ `Kimi k2.0` (Fallback).
- **Local Dept**: `Gemma 3:27b` (Primary) ➔ `Gemma 3:12b` (Fallback).

### Smart Memory & Safety
- **Context Memory**: Automatically summarizes conversation history every 10 exchanges, allowing for massive "infinite" logic retention without crashing the context window.
- **Auto-Backup**: The system automatically zips its own core engine files into the `/backups` folder periodically, ensuring your custom logic is never lost.
- **VRAM Management**: High-speed "Purge" logic for local models, optimized for AMD RX 7900 XTX (ROCm) and NVIDIA alike.
- **Visual Heartbeat**: Real-time "Neural Bridge: ACTIVE" status in the GUI header with automatic startup confirmation log.

---
*Note: This is a sovereign AI workstation. If you don't want to subscribe to Abacus.ai, you can still run the Local Gemma 3 models for FREE as no API_KEY is required. Enjoy the dual-engine power of the Quad!*