Basic Experimentation with LLM for Wireshark Packet Analysis

Overview

This project is an experimental demonstration of how a Large Language Model (LLM) can assist in analyzing network packets captured with Wireshark. It uses Python libraries and specialized models to summarize packet data and provide insights into potential anomalies.

Requirements

- Python 3.8+ - PyTorch - Transformers - Pyshark - LangChain - TikToken - OpenAI API Key

Getting Started

1. Install dependencies: ``` pip install -r requirements.txt ```
2. Add your OpenAI API key to an `.env` file as `OPENAI_API_KEY=YOUR_KEY`.
3. Run the Python scripts to capture packets, process summaries, or engage with the LLM-based chat interface.

How It Works

- Network packets are parsed for IP, port, protocol, and other headers. - An LLM or fine-tuned model processes packet data, generating embeddings or summaries. - A question-answering pipeline provides real-time insights into network behavior.

Contributing

Feel free to fork and modify this project. Submit pull requests for any interesting enhancements.