I built a URL threat detection system from scratch — here's what I learned from v0 to v2
A few months ago I posted here asking for advice and feedback on an early version of a phishing detector I was building. Today I'm sharing v2 — live, deployed, and significantly more powerful than where I started.
Here's the journey.
Where it started
The idea came from a real problem. People around me — friends, family, crypto community members — kept clicking suspicious links and getting scammed. There was no simple tool they could use to check a link before clicking. So I decided to build one.
v0 was basic. It worked but it was rough, limited, and I wasn't confident enough to share it widely. I actually hid it from my Facebook page because I was afraid of what people would say.
That fear wasted time. Lesson one.
What v2 became
v2 is a full threat detection system:
7 threat intelligence providers simultaneously — VirusTotal, Google Safe Browsing, PhishTank, ThreatFox, URLhaus, AbuseIPDB, DNS Blocklist
Custom ML model trained on 500,000+ real phishing and safe URLs
Risk score from 0 to 1 with full explanation
Deep scan mode for full site analysis
Bulk scanning for multiple URLs
Result caching for instant repeat scans
Results in under 20 seconds
What I learned
- Ship even when it's not ready v0 taught me more than any tutorial. Real feedback exposed problems I never would have caught alone.
- Provider redundancy matters When one provider is unavailable the system still works. Build for failure.
- ML alone is not enough My ML model sometimes flags legitimate URLs. The lesson — combine multiple signals, don't trust any single one.
- Transparency builds trust My DSA chairman reviewed the tool. His biggest concern wasn't detection — it was transparency. Users need to know what you check, how you score it, what you do with their data.
What's next
Privacy Policy and About page
Scoring methodology documentation
Private IP blocking to prevent abuse
Redis for persistent caching
Mobile app and browser extension
Try it
🔗 aeglixmind.com
Free. No login. No signup.
Feedback welcome — especially from security researchers and developers. Tell me what's wrong, what's missing, what could be better.
Built under Aeglix — cybersecurity tools for Africa. 🛡️
0 Comments
Log in to join the conversation.No comments yet. Be the first to share your thoughts.