Skip to content

Neural Shadow is a local-first, zero-knowledge desktop intelligence system. It acts as a photographic memory for your digital life, continuously capturing screen content and system audio, indexing it into a local vector database, and allowing for semantic natural language retrieval.

Notifications You must be signed in to change notification settings

Kshithijm1/NeuralShadow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Neural Shadow 🧠🌑

The AI that remembers everything you see and hear.

Neural Shadow is a local-first, zero-knowledge desktop intelligence system. It acts as a photographic memory for your digital life, continuously capturing screen content and system audio, indexing it into a local vector database, and allowing for semantic natural language retrieval.


🚀 Features

  • Total Recall: 24/7 Background capture of Screen (OCR) and Audio (Whisper).
  • Semnatic Search: Ask questions like "What command did I run for Docker last week?"
  • Diff-Based Optimization: Only saves screen updates when pixels change (>0.5%).
  • Local-First: 100% of data stays on your machine.
  • RAG Powered: Uses ChromaDB + Ollama (Mistral/Llama3) to answer questions contextually.
  • Encrypted: All raw screen/audio data is encrypted at rest using Fernet (AES).

🛠️ Architecture

The system is built on a modular "Cortex" architecture:

  1. Senses: mss (Screen) and soundcard (Audio) capture raw data.
  2. Cortex: OCR (Tesseract) and Whisper extract text.
  3. Memory: ChromaDB stores semantic embeddings (all-MiniLM-L6-v2).
  4. Gateway: FastAPI exposes a Search Endpoint.
  5. Interface: Next.js Cyberpunk Dashboard for interaction.

📦 Installation

Prerequisites

  • Python 3.10+
  • Node.js 18+
  • Tesseract OCR (Installed and added to PATH)
  • Ollama (Running ollama serve with mistral)

Quick Start (Windows)

  1. Clone the Repo
    git clone https://github.com/Kshithijm1/NeuralShadow.git
    cd NeuralShadow
  2. Install Backend
    pip install -r requirements.txt
  3. Install Frontend
    cd web
    npm install
  4. Run All Double-click start.bat or run:
    python main.py      # Terminal 1
    npm run dev --prefix web  # Terminal 2

💡 Usage Examples

Once the system is running, open http://localhost:3000 and press Cmd+K (or Ctrl+K) to open the search bar.

1. Developer Productivity

"What flags did I use for the docker build command yesterday?" "Show me the python script I was editing an hour ago."

2. Meeting Recall

"What action items were assigned to me during the sync?" "Summarize the feedback from the design review."

3. Digital Wellbeing

"What website was I looking at before I got distracted?"

🔒 Security

Neural Shadow is designed with privacy as the #1 priority.

  • Zero-Cloud: No data is sent to OpenAI or any cloud provider.
  • Encryption: By default, ENABLE_ENCRYPTION = True in core/config.py. All .png and .wav files are encrypted with secret.key.

🐳 Docker Support

Run the full stack in a container (Experimental):

docker-compose up --build

About

Neural Shadow is a local-first, zero-knowledge desktop intelligence system. It acts as a photographic memory for your digital life, continuously capturing screen content and system audio, indexing it into a local vector database, and allowing for semantic natural language retrieval.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published