Vexius Engine Documentation
Comprehensive guide to understanding, deploying, and building upon the Vexius Engine—a highly secure, AI-powered document collaboration platform.
Introduction
Vexius Engine is built to solve two major problems in modern enterprise collaboration: Data isolation and AI capability constraints.
By combining strict database-level Row-Level Security (RLS) with our proprietary Vexius Native Engine, Vexius ensures zero data leakage between workspaces without sacrificing performance. Furthermore, the built-in AI Copilot dynamically routes queries to the most optimal Vexius AI models based on task complexity and latency requirements.
Zero-Trust Architecture
Every document access is authenticated at the database level.
Real-time Sync
Multi-user editing powered by the Vexius Native Engine.
Architecture
Vexius is structured as a modern monorepo utilizing a frontend/backend split with a unified database schema.
- Frontend (Next.js): React Server Components, Framer Motion for animations, and Zustand for state management. Connects to Solana wallets for Web3 native authentication.
- Backend API (NestJS): A robust, modular monolith handling workspace creation, WOPI protocol endpoints, and AI prompt orchestration.
- Database (PostgreSQL via Supabase): The single source of truth. Uses Row-Level Security (RLS) to enforce workspace boundaries at the kernel level of the database.
- Vexius Editors: Proprietary, native components for Docs, Sheets, Slides, and PDF built directly into the frontend for maximum performance and seamless AI integration.
Authentication
Vexius supports seamless Web3 authentication. Instead of traditional email/password setups, users authenticate by cryptographically signing a message using their Solana wallet (e.g., Phantom, Solflare).
const signature = await wallet.signMessage(message);
// Send to backend for verification
const token = await api.post("/auth/verify", { publicKey, signature });
Workspaces & RLS
Everything in Vexius belongs to a Workspace. A workspace is an isolated logical container. To enforce this, we use PostgreSQL Row-Level Security.
Even if a vulnerability exists in the application code, the database itself will reject queries attempting to read documents from a workspace the user is not a member of.
CREATE POLICY "Users can read workspace documents"
ON documents FOR SELECT
USING (
workspace_id IN (
SELECT workspace_id FROM members WHERE user_id = auth.uid()
)
);
Vexius Native Engine
To provide enterprise-grade document editing, Vexius relies on its proprietary native engine, bypassing the need for third-party integrations like ONLYOFFICE or complex WOPI protocols.
Our suite includes four intelligent editors:
- Vexius Docs: A rich text collaborative editor with real-time AI drafting.
- Vexius Sheets: High-performance spreadsheets featuring AI-generated formulas and data analysis.
- Vexius Slides: Dynamic, layout-aware presentation generator utilizing responsive scaling.
- Vexius PDF: Secure document viewer with cryptographic signatures and on-chain verification capability.
Vexius Agent SDK
Vexius exports a dedicated Node.js library @vexius/sdk letting developers implement, validate, and execute customized agent routines outside the Vexius core module.
const agent = new VexiusAgent({
id: "custom-writer",
name: "Writer Agent",
version: "1.0.0",
permissions: ["workspace.read"]
});