PhishGuardAI Documentation

Complete guide to using our AI-powered phishing detection tool

Table of Contents

1. Overview2. How It Works3. Getting Started4. Features5. API Reference6. Technical Details7. FAQ8. Troubleshooting

1. Overview

PhishGuardAI is an advanced phishing detection tool that uses machine learning algorithms to analyze URLs and determine if they're potentially malicious. Built as a final year research project, it combines cutting-edge AI technology with user-friendly design to protect users from phishing attacks.

Instant Detection

Get results in seconds

90% Accuracy

Highly accurate ML model

Free to Use

No registration required

2. How It Works

1

URL Input

User enters a URL they want to check for phishing indicators

2

Feature Extraction

Our system extracts 14+ features from the URL including domain age, SSL certificate status, URL structure, and more

3

ML Analysis

The machine learning model analyzes these features using trained algorithms to detect phishing patterns

4

Results

Get instant results with confidence score and detailed analysis

3. Getting Started

Quick Start

  1. Visit the PhishGuardAI homepage
  2. Copy the URL you want to check
  3. Paste it into the input field
  4. Click "Check URL" or press Enter
  5. View your results instantly

⚠️ Important Notes

  • • Always include the full URL (with http:// or https://)
  • • Results are for informational purposes only
  • • When in doubt, avoid clicking suspicious links

4. Features

Detection Capabilities

  • ✓URL structure analysis
  • ✓Domain reputation checking
  • ✓SSL certificate validation
  • ✓Suspicious pattern detection
  • ✓Typosquatting identification

User Experience

  • ✓Instant results (2-3 seconds)
  • ✓Confidence score display
  • ✓Detailed analysis breakdown
  • ✓Mobile-friendly interface
  • ✓No registration required

5. API Reference

Endpoint

POST - https://phishguardaibackend-production-e66e.up.railway.app/predict

Request Body

{
  "url": "https://example.com"
}

Response

{
    "URL": "url": "https://example.com",
    "Prediction": "Phishing",
    "Raw_Prediction_Value": 0,
    "Risk_Level": "High Risk",
    "Domain_Age": "14083 days",
    "Network_Information": {
        "IP Address": "18.155.68.38",
        "Location": "Unknown",
        "Hostname": "new.express.adobe.com",
        "Response Time": "2.25 ms",
        "Subdomain Count": 2,
        "Redirection Count": 0,
        "Redirection Chain": [
            "https://new.express.adobe.com/webpage/TBPcFvEjDVKw9"
        ]
    },
    "SSL_Certificate": {
        "Status": "Valid",
        "Expiry Date": "2026-05-06",
        "Issuer": "Amazon"
    },
    "Features": {
        "having_IP_Address": 0,
        "URL_Length": 51,
        "Shortining_Service": 0,
        "having_At_Symbol": 0,
        "double_slash_redirecting": 0,
        "Prefix_Suffix": 0,
        "SSLfinal_State": 0,
        "having_Sub_Domain": 1,
        "Domain_registeration_length": 14425,
        "Request_URL": 0.0,
        "URL_of_Anchor": 0.0,
        "Links_in_tags": 0.6521738846880919,
        "SFH": 1
    },
    "Probability_Legitimate": "25.38%",
    "Probability_Phishing": "74.62%"
}

6. Technical Details

Machine Learning Model

Our model is trained on a comprehensive dataset of known phishing and legitimate websites, using advanced feature engineering and ensemble methods.

Algorithms: Decision Tree Classifier | Random Forest | Gradient Boosting | Naïve Bayes | Weighted Soft Voting
Training Data: 100,000+ URLs
Features: 14+ extracted features
Accuracy: 89.2% on test set

Key Features Analyzed

  • • URL length and complexity
  • • Domain registration age
  • • SSL certificate status
  • • Number of subdomains
  • • Presence of IP addresses
  • • Suspicious keywords
  • • URL shorteners usage
  • • Special characters count
  • • Domain reputation score
  • • Redirect chains
  • • WHOIS information
  • • Brand impersonation patterns

7. Frequently Asked Questions

How accurate is PhishGuardAI?

Our machine learning model achieves 95.2% accuracy on our test dataset, which includes thousands of verified phishing and legitimate websites.

Is PhishGuardAI free to use?

Yes, PhishGuardAI is completely free to use. There are no registration requirements or usage limits.

What types of URLs can I check?

You can check any HTTP or HTTPS URL. This includes websites, email links, social media links, and shortened URLs.

How long does analysis take?

Most URL analyses complete within 2-3 seconds. Complex URLs with multiple redirects may take slightly longer.

Do you store the URLs I check?

We do not store or log the URLs you submit for analysis. All processing is done in real-time and data is not retained.

Can I integrate PhishGuardAI into my application?

Currently, this is a research project and not available for commercial integration. However, you can use our web interface for personal use.

8. Troubleshooting

Common Issues

Error: "Invalid URL format"

Make sure your URL includes http:// or https:// at the beginning.

Error: "Analysis timeout"

This may occur with very slow-loading websites. Try again in a few moments.

Unexpected results

Remember that no detection system is 100% perfect. Use results as guidance alongside your own judgment.

Best Practices

  • • Always verify suspicious results manually
  • • Check multiple sources for important decisions
  • • Report false positives to help improve the system
  • • Keep your browser and security software updated

Need More Help?

This is a research project developed as part of a final year study. For technical questions or feedback, please contact our development team.

PhishGuardAI - Final Year Research Project

Built with Next.js and Machine Learning