PhishGuard — AI-Powered Gmail Phishing Detector

A Chrome browser extension that detects phishing emails in Gmail using a Naive Bayes ML model served via a Python Flask backend.

---

🚀 How It Works

1. Auto-Scan: When you open an email in Gmail, the extension silently extracts the email text.
2. Cloud Analysis: It sends the text to the live PhishGuard Flask API hosted on Render.
3. ML Prediction: The backend uses a trained Multinomial Naive Bayes model with TF-IDF vectorization to determine if the email is safe or a phishing attempt.
4. Warning Banner: A banner drops down in Gmail (Red for Phishing, Green for Safe) showing the AI's confidence score.

---

📁 Project Structure

phishguard-backend/
├── app.py                # Flask REST API server (Production-ready via Gunicorn)
├── train_model.py        # ML training script (Naive Bayes + TF-IDF)
├── requirements.txt      # Python dependencies
└── model/                # Pre-trained models loaded by app.py
    ├── phishing_model.pkl
    └── tfidf_vectorizer.pkl

(Note: The Chrome Extension code is maintained in a separate directory/repository.)

---

☁️ Cloud Deployment (Render)

This backend is designed to be hosted for free on Render as a Web Service.

Render Setup:

• Environment: Python 3
• Build Command: pip install -r requirements.txt.
• Start Command: gunicorn app: app (This is critical! It ensures the ML models load into memory at startup)

---

💻 Local Setup & Training

If you want to train your own model or run the API locally:

1. Install Dependencies

pip install -r requirements.txt

2. Train the Model

python train_model.py

This script downloads a dataset of ~18,000 phishing/safe emails from Kaggle, cleans the text (removes HTML, links, stop words), and trains the Naive Bayes classifier, saving the output to the model/ folder.

3. Run Locally

python app.py

The server will start at http://localhost:5000.

4. API Endpoints

Endpoint	Method	Description
/api/health	GET	Server status & model check
/api/predict	POST	Analyze single email
/api/batch	POST	Analyze multiple emails

---

⚙️ Tech Stack

• ML Model: Scikit-learn Multinomial Naive Bayes
• Text Processing: NLTK (PorterStemmer, Stopwords), TF-IDF (10,000 features, bigrams)
• Backend API: Python Flask + Flask-CORS + Gunicorn
• Deployment: Render (Free Tier)
• Frontend Extension: Vanilla JS/CSS (Manifest V3)