⚠️ EDUCATIONAL USE ONLY — AUTHORIZED TESTING ONLY. This project exists for education, research, and defense of systems you own or hold explicit written authorization to assess. Unauthorized use is prohibited and may be illegal. Read ETHICS.md and SCOPE.md before use. Use at your own risk; AS IS, no warranty.

A Secure Warning Platform Against Web Attacks Using Machine Learning

A machine-learning educational research project that detects phishing URLs, Denial-of-Service (DoS) traffic, and cross-site scripting (XSS) payloads with scikit-learn classifiers, Jupyter notebooks, and Flask demo apps.

Machine-learning based web-attack warning platform covering Phishing, DoS and XSS detection with trained classifiers, feature engineering, and a front-end demonstration.

Why this project

Web attacks are increasingly automated, and ML offers a way to generalize detection beyond fixed signatures. This repository is a learning-oriented implementation of that idea: it applies classic and ensemble classifiers to well-known datasets — a URL dataset for phishing detection, the NSL-KDD dataset for intrusion/DoS detection, and an XSS payload corpus for script detection — then wraps the trained models in simple Flask demos. Studying the notebooks teaches feature extraction, model comparison, and accuracy measurement on real security datasets. Everything here is for educational and authorized use (e.g., your own lab web app, academic coursework).

Features

• Phishing URL detection — Phishing URL Detection Flask app with gradient-boosting classifier and a 16-feature URL extractor (feature.py), trained on urldata.csv, served by app.py.
• DoS / intrusion detection — DoS Attack (Intrusion Detection System) Flask app using the NSL KDD dataset (feature engineering via dummies and outlier capping) and a persisted model.pkl to classify NORMAL / DOS / PROBE / R2L / U2R.
• XSS detection — XSS notebook and script (XSS.PY) comparing LogisticRegression, RandomForest, SVC, GaussianNB, and a Keras network over a CountVectorizer bag-of-words on XSS_dataset.csv.
• Notebooks — step-by-step Jupyter walkthroughs for every detector, including pandas_profiling.html for DoS data exploration.
• Demo front-end — Front-End Code/index.html landing page tying the project together.
• Deployable Flask apps — Procfile (Gunicorn) and a Dockerfile for the intrusion-detection app.

Quickstart

Prerequisites: Python 3, scikit-learn, pandas, numpy, flask, jupyter.

# Explore the training and evaluation notebooks
jupyter notebook "Phishing Attack/Phishing-URL-Detection-master/Phishing-URL-Detection-master/Phishing URL Detection.ipynb"
jupyter notebook "DoS Attack (Intrusion Detection System)/Network Intrusion Detection System (DoS Attack)/Network Intrusion Detection System.ipynb"
jupyter notebook "XSS (Cross Site Scripting)/XSStrike-master/XSS_dataset.csv (1)/XSS.ipynb"

# Run the phishing detection web app (data + model artifacts are in the folder)
cd "Phishing Attack/Phishing-URL-Detection-master/Phishing-URL-Detection-master"
python app.py

# Run the DoS/intrusion detection web app (model.pkl is committed)
cd "../../../DoS Attack (Intrusion Detection System)/Network Intrusion Detection System (DoS Attack)"
python app.py

Project structure

Phishing Attack/.../    phishing URL detector: feature.py, app.py, urldata.csv
DoS Attack (Intrusion Detection System)/...  NSL-KDD detector: app.py, model.pkl, notebooks
XSS (Cross Site Scripting)/...   XSS classifier: XSS.PY, XSS_dataset.csv, notebook
Front-End Code/         static landing page (index.html, style.css, images)

Documentation

• ETHICS.md and SCOPE.md — authorized-use rules for this educational project.
• SECURITY.md — responsible disclosure.
• CONTRIBUTING.md / CODE_OF_CONDUCT.md — community guidelines.

Notes

The dataset paths inside the notebooks and app.py are machine-specific (hardcoded Windows paths in places) — update them to your local checkout. The intrusion-detection Dockerfile expects a requirements.txt at build time; add one with Flask, scikit-learn, pandas, numpy, and joblib before building.

License

MIT. Educational deliverable — run it only against your own data, your own lab, or data you are authorized to analyze.