linksentry 0.1.0

A CLI tool to detect phishing URLs using machine learning

LinkSentry

A command-line tool to detect phishing URLs using machine learning.

Features

• 🔍 Single URL Check - Quickly analyze any URL for phishing indicators
• 📁 Batch Processing - Check multiple URLs from a file
• 🤖 ML-Powered - Uses Random Forest trained on 88,000+ URLs with 111 features
• 🌐 Optional Deep Analysis - DNS, WHOIS, and SSL lookups with --full flag
• 📊 Multiple Output Formats - Plain text or JSON output

Installation

pip install linksentry

For full feature extraction (DNS/WHOIS lookups):

pip install linksentry[full]

Usage

Check a Single URL

# Basic check
linksentry check "https://suspicious-site.xyz/login"

# With full feature extraction (DNS/WHOIS)
linksentry check "https://example.com" --full

# Output as JSON
linksentry check "https://example.com" --json

Check Multiple URLs

# Check URLs from a file (one URL per line)
linksentry check-file urls.txt

# Save results to CSV
linksentry check-file urls.txt --output results.csv

# With JSON output
linksentry check-file urls.txt --json

Train Custom Model

# Train with your own dataset
linksentry train --data your_dataset.csv

# Specify output path
linksentry train --data dataset.csv --output custom_model.pkl

Show Model Info

linksentry info
linksentry info --json

Example Output

$ linksentry check "https://secure-login.suspicious-domain.xyz/verify"

⚠️  PHISHING
URL: https://secure-login.suspicious-domain.xyz/verify
Confidence: 94.2%
P(Legitimate): 0.0583
P(Phishing): 0.9417

$ linksentry check "https://github.com/user/repo"

✅ LEGITIMATE
URL: https://github.com/user/repo
Confidence: 98.7%
P(Legitimate): 0.9871
P(Phishing): 0.0129

Exit Codes

Code	Meaning
0	URL is legitimate
1	URL is phishing (or at least one phishing URL in batch)
2	Error occurred

This allows integration with shell scripts:

if linksentry check "$URL" > /dev/null 2>&1; then
    echo "URL is safe"
else
    echo "URL may be phishing!"
fi

Feature Extraction

LinkSentry extracts 111 features from URLs:

Category	Features	Description
URL Characters	18	Count of special characters in full URL
Domain	21	Domain length, characters, IP check
Directory	18	Path analysis
File	17	Filename analysis
Parameters	20	Query string analysis
External*	17	DNS, WHOIS, SSL, response time

*External features require --full flag and linksentry[full] installation.

Dataset Format

For training custom models, your CSV should have:

• 111 feature columns (see feature list above)
• 1 target column named phishing (0 = legitimate, 1 = phishing)

Python API

from linksentry import predict_url, extract_features

# Predict a URL
result = predict_url("https://example.com")
print(result['label'])  # 'legitimate' or 'phishing'
print(result['confidence'])  # 0.0 to 1.0

# Extract features only
features = extract_features("https://example.com", full=True)

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.