gresecml 0.3.5

A small CLI security toolkit

GresecML

This command-line tool analyzes network traffic from either a .pcap file or a live capture, and makes predictions using a TensorFlow model. Results can be displayed in the console and/or written to an HTML/CSV file.

---

⚠️ Important

This tool was developed solely as part of an academic school project. It is intended for educational purposes only and should not be used in production environments or relied upon for professional applications.

Do not use this tool to scan network traffic unless you own it or have proper authorization.

---

Features

• Analyze .pcap files or capture of live network traffic
• Predict traffic sessions using a TensorFlow model
• Export results to an HTML report
• Filter sessions by probability thresholds
• Support for lazy loading to save memory
• Verbose mode for detailed console output

---

Workflow

The prediction pipeline follows these steps:

1. Capture – Collects packets from a file or live network interface
2. Sessions – Groups packets into sessions
3. Prediction – Runs sessions through the TensorFlow model
4. Output – Displays results in console and/or export to HTML

---

Installation

Install with pip:

pip install gresecml

---

Usage

Run the CLI with:

gresecml [OPTIONS] COMMAND [ARGS]...

For help:

gresecml --help

---

Examples with "gresecml tf predict":

• Analyze a .pcap file with tensorflow and save results to HTML

  gresecml tf predict -i traffic.pcap -o results.html
  

• Run live capture on default interface for 60 seconds

  gresecml tf predict
  

• Run live capture on a specific interface with custom timeout

  gresecml tf predict -if eth0 -t 120
  

• Enable verbose output and full HTML report

  gresecml tf predict -i traffic.pcap -o results.html -v -efo
  

• Filter sessions with normal probability ≤ 70%

  gresecml tf predict -i traffic.pcap -pnm 70
  

• Use lazy loading to save memory

  gresecml tf predict -i traffic.pcap -ll
  

Notes

• If no --input is provided, the tool defaults to live capture.
• Lazy loading is recommended for large .pcap files to reduce memory usage.
• The HTML output provides a structured table of predictions for possible further investigation. The table is sorted by the prediction_normal column.

Example Output

When running with --verbose, predictions will be printed in the console.
If --output is specified, results will also be saved as an HTML file.

---