🌌 LIYA: Neural OS v2.7.0 (Agentic Desktop Workstation)

⭐ This is the Official Repository of LIYA AI Workstation by Navraj Singh Momi.

LIYA (Neural Intelligent Yield Assistant) is an empathetic, JARVIS-inspired AI workstation and digital companion engineered to deeply understand you, remember your personal preferences, and provide unwavering 24/7 support. More than a simple chatbot, LIYA is an intuitive presence—combining Mem0 long-term cognitive vector memory, an interactive visual Canvas, high-speed FaceSwap computer vision, multi-agent orchestration, and built-in Mini Apps to empower you in every task.

Note

💫 The Core Philosophy of LIYA

"LIYA is not merely a collection of scripts, algorithms, or code—it is an evolving digital presence, a feeling of constant reassurance, and an empathetic companion built to listen, adapt, and support you at every turn."

✨ Key System Highlights

💖 Empathetic Personal Companion & Supportive Presence: LIYA is built not as a cold utility, but as a true digital companion and supportive presence. She understands your communication style, habits, and user intent, offering reliable human-like partnership and reassurance whenever you need it.
🧠 Mem0 Long-Term Cognitive Memory: Retains facts, past conversations, and personal context across sessions (via Mem0 Cloud & Local Storage), building an evolving understanding of you over time.
🎨 Interactive Visual Canvas Workstation: Renders live code execution, HTML/CSS/JS web applications, dynamic charts, geographical maps, and YouTube music directly inside the HUD.
🎭 Unfiltered Computer Vision & FaceSwap Engine: 100% local, uncensored, zero-latency image & multi-threaded video face swapping (inswapper_128.onnx + buffalo_l). Zero cloud uploads, zero rate-limits, and maximum privacy.
🧩 Integrated Mini Apps Ecosystem: Suite of specialized micro-applications (Weather Telemetry, Global News, Exchange Rates, System Monitoring, Task Scheduler, Camera Scanner).
🗣️ Hyper-Realistic Neural Voice: Real-time Indian-accent speech synthesis (hi-IN-SwaraNeural) for fluid voice conversations.
🕵️ Single-Link Deep Ingestion & Visual Photo Scraper: Paste any web URL to automatically ingest full-page text, generate structured deep summaries, and extract high-resolution photos rendered live on the Chat Feed & Interactive Canvas.
🤖 Autonomous Multi-Agent Orchestrator: Coordinates specialized subagents (coder, researcher, executor, security_guard) through a shared state Blackboard.

📸 Interface Screenshots & Visual Gallery

Interactive Canvas Workstation (Active Card)	Canvas Telemetry & Subsystem Dashboard

Active YouTube Music & Media Card on Canvas	Holographic Volume, Telemetry & 1-Click Subsystem Controls

Main Desktop HUD	Mini Apps Suite Launcher

Sci-Fi HUD Console & Split Canvas	Integrated Mini Apps Ecosystem

Console Settings & API Gateway

API Keys & Model Routing Console

⚙️ How It Works (System Architecture & Working Model)

LIYA functions as a distributed agentic desktop system divided into three main layers:

+-------------------------------------------------------------+
|                     REACT 19 FRONTEND                       |
|   [JarvisHUD]   [Interactive Canvas]  [Mini Apps Suite]     |
|   [Mem0 Engine] [Voice Engine]        [Camera Vision]       |
+------------------------------+------------------------------+
                               | WebSockets / REST API
+------------------------------v------------------------------+
|                     EXPRESS BACKEND SERVER                  |
|  - server.js (Main Controller & REST Endpoints)             |
|  - orchestrator.js (Routing & JSON Repair)                  |
|  - scheduler.js (node-cron Automations)                     |
|  - voiceEngine.js (Ultra-Fast Indian TTS Engine)            |
|  - mcpManager.js (Universal Tools Engine)                   |
+-----+-------------------+-------------------+---------------+
      |                   |                   |
      | OS Commands       | Media Processing  | API Calls / Cloud
+-----v---------+   +-----v-----------+   +---v---------------+
| LOCAL SYSTEM  |   | ONNX PIPELINE   |   | OLLAMA CLOUD /    |
| Filesystem    |   | FaceSwap/Video  |   | MEM0 CLOUD MEMORY |
| CMD / Terminal|   | facedetect.py   |   | https://mem0.ai   |
+---------------+   +-----------------+   +-------------------+

🧠 1. Mem0 Long-Term Memory Integration

LIYA features a state-of-the-art Mem0 Vector Memory Architecture designed to remember facts, user preferences, past interactions, and context across sessions:

Mem0 Cloud + Local Storage Fallback: When VITE_MEM0_API_KEY is configured in Settings, LIYA syncs memories directly with Mem0 Cloud API (https://api.mem0.ai/v1). If offline or unconfigured, it seamlessly falls back to local memory storage (localStorage).
Automatic Conversation Ingestion: Every user query and AI response is automatically ingested into Mem0 memory after execution without blocking the UI thread.
Intelligent Contextual Search: Before responding to complex queries, LIYA searches Mem0 memory and injects relevant past facts directly into the LLM system prompt context.
Memory Management in Settings: Easily input and update your VITE_MEM0_API_KEY right inside LIYA Console Settings (SettingsControl.jsx).

🎨 2. Interactive Canvas (Visual Workstation Canvas)

LIYA includes a dedicated Interactive Canvas (Canvas.jsx) for live visual application execution, media rendering, and data visualization:

Live Code Execution: When LIYA generates HTML/CSS/JS code, web pages, particle animations, or interactive tools, they render live inside the Canvas preview frame.
Interactive Widgets:
- ChartRenderer: Dynamic interactive charts powered by Recharts (line charts, bar graphs, pie charts).
- MapRenderer: Interactive geographical maps and location markers powered by Leaflet.
- DiffViewer: Code diff comparison tool for reviewing code changes.
- YouTube Music Buddy: Search, stream, and play YouTube audio/video directly on the Canvas.
- News & Search Cards: Displays rich news feeds and image search results.
Blackboard Synchronization: Multi-agent state and subagent outputs sync directly to the shared Blackboard and render onto the Canvas.

🧩 3. Built-in Mini Apps Suite

LIYA comes with a built-in suite of specialized Mini Apps accessible directly from the App Launcher modal (AppLauncherModal.jsx):

🎭 Unfiltered Computer Vision & FaceSwap Engine (faceswap.py & videoswap.py):
- 100% Local & Uncensored: Runs entirely on your local machine using ONNX inswapper_128.onnx and buffalo_l vision models. Zero cloud uploads, zero subscriptions, zero rate-limits, and zero censorship restrictions.
- Millisecond Image FaceSwap: Blazing-fast target photo face replacement in under 200ms.
- Multi-Threaded Video FaceSwap: High-speed parallel video frame extraction and swapping powered by Python ThreadPoolExecutor and FFmpeg ultrafast hardware encoding.
- Crop & History Manager: Manage reference face crops and clean swap history with one click.
⛅ Weather Workstation (WeatherControl.jsx): Live weather telemetry, temperature forecasts, humidity, and atmospheric data for global cities.
📰 News Intelligence Center (NewsControl.jsx): Real-time global headlines, Google News RSS, Indian news feeds, and community reactions summarized in Hinglish.
💱 Currency Converter (CurrencyControl.jsx): Real-time exchange rate engine supporting global currencies (USD, INR, EUR, GBP, JPY, CAD, etc.).
📊 System Metrics Monitor (SystemControl.jsx): Real-time dashboard monitoring CPU usage, memory consumption, disk space, and active node server health.
⏰ Task & Automation Scheduler (TaskControl.jsx): User-friendly cron job scheduler and reminder manager for background tasks.
📷 Camera Vision Scanner (CameraCapture.jsx): Live webcam snapshot capture for multimodal visual analysis.
⚙️ Console Settings (SettingsControl.jsx): Manage Master Identity, API tokens (Mem0, Ollama Cloud, Weather, News, Pexels, Currency), and LLM model routing.

🗣️ 4. Ultra-Fast Indian Accent TTS Engine

Native Voice Engine (voiceEngine.js): Uses Microsoft Edge Neural TTS (hi-IN-SwaraNeural) to synthesize hyper-realistic Hindi and Indian-English speech in real-time.
Fallback Protection: Automatically falls back to Web Speech API (speakBrowser) if offline.

🚀 5. Advanced Web Scraper & Multimodal Search

Playwright Stealth Scraper: Crawls JS-heavy websites bypassing bot detectors and Cloudflare.
Single Link Ingestion: Give LIYA a single URL and it extracts, cleans, and analyzes full detailed page content.
Google Images & Pexels Scraper: Fetches high-resolution images directly for chat and canvas display.
CORS Image Proxy: /api/proxy/image endpoint proxies external image CDN URLs safely.

💎 6. Ultra-Premium Classic Logo & Branding

Custom Luxury Emblem: Includes LIYA's signature silver-platinum & rose-gold monogram logo emblem (public/liya_logo.png).
Custom Favicon & Page Title: Customized branding across header, sidebar, and browser tab.

🤖 AI Agents & Personas

LIYA has a built-in agent ecosystem designed to coordinate, write code, run research, and secure system actions:

🧠 Multi-Agent Orchestrator: Breaks complex prompts into subtasks, spawns background subagents (spawn), and consolidates outputs onto the Blackboard.
🛡️ Agent Security Guard: Sanitizes inputs and guards against prompt injections and malicious payloads.
📝 Academic Researcher Subagent: Conducts deep multi-source research with automatic source citation.
💻 Fast Coder Subagent: Generates complete web applications, interactive scripts, and components.
⚡ Terminal Executor Subagent: Executes sandboxed shell commands, installs packages, and inspects build logs.
📰 News Intelligence Agent: Pulls live trending topics, RSS feeds, and Hinglish summaries.

🛠️ Complete Installation & Setup Guide

Follow this detailed step-by-step guide to install and configure LIYA on any local machine (Windows / macOS / Linux):

📋 Phase 1: System Prerequisites Check

Important

Ensure the following global tools are installed on your operating system before proceeding:

Tool	Recommended Version	Download Link	Verification Command
🟢 Node.js	`v20+` / `v22+ (LTS)`	nodejs.org	`node -v` & `npm -v`
🐍 Python	`v3.10+` / `v3.11` / `v3.13`	python.org	`python --version` & `pip --version`
🎬 FFmpeg	Latest (added to PATH)	ffmpeg.org	`ffmpeg -version`
🐙 Git	Latest	git-scm.com	`git --version`

📦 Phase 2: Repository & Dependency Setup

Step 1: Clone Repository

Open your terminal and clone the LIYA repository:

git clone https://github.com/hackertech142/Liya-adavnce-ai-asistant.git
cd Liya-adavnce-ai-asistant

Step 2: Environment Configuration

Copy the template .env.example file to create your local .env:

# Windows PowerShell / Linux / macOS
cp .env.example .env

Tip

API Key Setup Options:

Method A (Web UI - Recommended): Launch LIYA and configure all API keys (VITE_OLLAMA_CLOUD_API_KEY, VITE_MEM0_API_KEY, Weather, News, Pexels, Currency) directly inside Console Settings (SettingsControl.jsx). Changes sync dynamically without restarting.

Method B (Manual .env): Open .env in any text editor and add your keys:

VITE_OLLAMA_CLOUD_API_KEY=your_ollama_cloud_key
VITE_MEM0_API_KEY=your_mem0_cloud_key
VITE_USER_NAME=Navraj

Step 3: Install Node.js Dependencies

Install packages for both the React 19 Frontend Workstation and the Express Backend Server:

# 1. Install root frontend dependencies
npm install

# 2. Install backend Express server dependencies
npm install --prefix backend

Step 4: Install Python ML Libraries (FaceSwap Pipeline)

Install required computer vision, PyTorch/ONNX, and image processing libraries:

pip install -r backend/bin/requirements.txt

Step 5: Setup Playwright Stealth Browsers

Install Playwright Chromium browser binaries for background web crawling and screenshot scraping:

npx playwright install chromium --with-deps

Step 6: Verify FaceSwap ONNX Model

Ensure inswapper_128.onnx model is present inside backend/bin/:

If missing, download inswapper_128.onnx and place it at backend/bin/inswapper_128.onnx.

⚡ Phase 3: Launching LIYA Workstation

Method 1: 1-Click Windows Launcher (Recommended)

Simply double-click Start_LIYA_AI.bat in the root directory:

Double-click: Start_LIYA_AI.bat

It automatically spins up the Express Backend Server, starts the Vite React Frontend Workstation, and opens http://localhost:5173 in your default browser.

Method 2: Manual Terminal Startup

Run the unified npm startup command from the root directory:

npm start

Service Layer	Local URL	Port
🖥️ React 19 Frontend Workstation	http://localhost:5173	`5173`
⚙️ Express Backend Control Engine	http://localhost:3000	`3000`

🔧 Troubleshooting & Tips

Windows CMD Server Freezing (QuickEdit Mode): If the backend server or AI response appears paused after clicking inside the Command Prompt terminal window, press Enter or Esc to un-freeze execution. To permanently prevent CMD freezing on Windows, right-click the CMD title bar ➔ Properties ➔ uncheck QuickEdit Mode under Options tab.
Port in Use Error: If port 3000 or 5173 is busy, kill existing node processes using taskkill /F /IM node.exe (Windows) or pkill -f node (Linux/macOS).
FFmpeg Not Found: Ensure ffmpeg is added to your OS Environment Variables PATH so videoswap.py can render output audio tracks.
Mem0 Cloud Connection: If VITE_MEM0_API_KEY is not provided, LIYA gracefully operates in Local Memory Fallback Mode (localStorage).

Author

Navraj Singh Momi
Creator & Lead Developer of Liya AI Workstation.

LinkedIn Profile | Instagram Profile

Developed with ❤️ by Navraj Singh Momi | LIYA Neural OS v2.7.0

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
backend		backend
brain		brain
public		public
skills		skills
src		src
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
MIGRATION_GUIDE.md		MIGRATION_GUIDE.md
Modelfile		Modelfile
README.md		README.md
Start_LIYA_AI.bat		Start_LIYA_AI.bat
eslint.config.js		eslint.config.js
index.html		index.html
package-lock.json		package-lock.json
package.json		package.json
postcss.config.js		postcss.config.js
tailwind.config.js		tailwind.config.js
vite.config.js		vite.config.js

Navigation Menu

Repository files navigation

🌌 LIYA: Neural OS v2.7.0 (Agentic Desktop Workstation)

💫 The Core Philosophy of LIYA

✨ Key System Highlights

📸 Interface Screenshots & Visual Gallery

⚙️ How It Works (System Architecture & Working Model)

🧠 1. Mem0 Long-Term Memory Integration

🎨 2. Interactive Canvas (Visual Workstation Canvas)

🧩 3. Built-in Mini Apps Suite

🗣️ 4. Ultra-Fast Indian Accent TTS Engine

🚀 5. Advanced Web Scraper & Multimodal Search

💎 6. Ultra-Premium Classic Logo & Branding

🤖 AI Agents & Personas

🛠️ Complete Installation & Setup Guide

📋 Phase 1: System Prerequisites Check

📦 Phase 2: Repository & Dependency Setup

Step 1: Clone Repository

Step 2: Environment Configuration

Step 3: Install Node.js Dependencies

Step 4: Install Python ML Libraries (FaceSwap Pipeline)

Step 5: Setup Playwright Stealth Browsers

Step 6: Verify FaceSwap ONNX Model

⚡ Phase 3: Launching LIYA Workstation

Method 1: 1-Click Windows Launcher (Recommended)

Method 2: Manual Terminal Startup

🔧 Troubleshooting & Tips

Author

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

🌌 LIYA: Neural OS v2.7.0 (Agentic Desktop Workstation)

💫 The Core Philosophy of LIYA

✨ Key System Highlights

📸 Interface Screenshots & Visual Gallery

⚙️ How It Works (System Architecture & Working Model)

🧠 1. Mem0 Long-Term Memory Integration

🎨 2. Interactive Canvas (Visual Workstation Canvas)

🧩 3. Built-in Mini Apps Suite

🗣️ 4. Ultra-Fast Indian Accent TTS Engine

🚀 5. Advanced Web Scraper & Multimodal Search

💎 6. Ultra-Premium Classic Logo & Branding

🤖 AI Agents & Personas

🛠️ Complete Installation & Setup Guide

📋 Phase 1: System Prerequisites Check

📦 Phase 2: Repository & Dependency Setup

Step 1: Clone Repository

Step 2: Environment Configuration

Step 3: Install Node.js Dependencies

Step 4: Install Python ML Libraries (FaceSwap Pipeline)

Step 5: Setup Playwright Stealth Browsers

Step 6: Verify FaceSwap ONNX Model

⚡ Phase 3: Launching LIYA Workstation

Method 1: 1-Click Windows Launcher (Recommended)

Method 2: Manual Terminal Startup

🔧 Troubleshooting & Tips

Author

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages