Local Cybersecurity Assistant & RAG Platform

Cyber_LLM is a specialized, privacy-focused Large Language Model (LLM) interface designed for cybersecurity professionals. It features a local RAG (Retrieval-Augmented Generation) pipeline, allowing users to analyze sensitive documents (logs, reports, PCAPs converted to CSV) without data leaving their local machine.

🚀 Key Features

• Local RAG Pipeline: Upload PDF, CSV, TXT, or JSON files. The system chunks, embeds, and stores them in a local Vector DB (ChromaDB) for context-aware answers.
• Cybersecurity Specialized: Includes custom system prompts for Threat Analysis, IOC extraction, and Executive Summaries.
• Multi-Model Support: Switch between GGUF (Llama.cpp) and HuggingFace models (e.g., Qwen, DeepSeek, Lily).
• Visual Analysis: Support for Vision-Language models (OCR and image analysis).
• Modern UI: A responsive React-based chat interface with citation tracking and source highlighting.

---

🛠️ Tech Stack

Frontend

• Framework: React 18 + Vite
• Language: TypeScript
• Styling: CSS Modules + Lucide React Icons
• State: React Hooks

Backend

• API: FastAPI (Python 3.12+)
• Vector DB: ChromaDB (Persistent local storage)
• LLM Engine: llama-cpp-python (GGUF) & transformers
• Orchestration: Custom Service/Factory pattern

---

⚡ Quick Start Guide

Prerequisites

• Python: 3.12 or higher
• Node.js: 18 or higher
• Hardware: GPU recommended (NVIDIA CUDA) for LLM inference, though CPU execution is supported for GGUF models.

1. Backend Setup

cd backend

# Create and activate virtual environment
python -m venv .venv
source .venv/bin/activate  # Linux/Mac
# .venv\Scripts\activate   # Windows

# Install dependencies
pip install -r requirements.txt

# Run the API server
uvicorn main:app --reload --host 0.0.0.0 --port 8000

4. Frontend Setup

npm install 

npm run dev

5. Docker Setup

Frontend: Docker-compose up --build frontend

Backend: Docker-compose up --build backend

Both: Docker-compose up --build