PhishShield – Phishing Link Detector

Inspiration

Phishing is the most common cyber attack in the world. Fake bank, UPI, and delivery-scam links trick students, parents, and elders every day. I wanted a simple tool anyone can use to check a link before clicking it.

What it does

PhishShield takes a URL, analyses it, and returns a verdict (Safe / Suspicious / Phishing) along with a plain-language explanation of why.

How I built it

1. Feature extraction: URL length, number of dots and hyphens, "@" symbol, IP address instead of a domain, HTTPS usage, suspicious keywords (login, verify, update), and domain age.
2. Rule-based checks: quick red flags such as lookalike domains (e.g. paypa1.com).
3. Machine learning model: a Random Forest classifier trained on a public phishing URL dataset.
4. Web interface: a simple Flask app where the user pastes a link and gets an instant result.

Challenges I faced

• Finding a clean, balanced dataset
• Reducing false positives on legitimate long URLs
• Explaining the result in simple language for non-technical users

What I learned

• How phishing attacks are structured
• Feature engineering for security data
• Building and deploying a small ML-powered web app

What's next

• A browser extension for real-time warnings
• Tamil and other regional-language alerts
• QR-code scanning to catch "quishing"

Built With

• bootstrap
• css3
• cybersecurity
• flask
• git
• github
• google
• html5
• javascript
• joblib
• jupyter-notebook
• machine-learning
• matplotlib
• numpy
• pandas
• phishing-detection
• python
• python-whois
• random-forest
• regex
• requests
• rest-api
• scikit-learn
• tldextract
• vs-code