Scam Shield

We’ve spent months building a tool that doesn't just say "this looks fishy"—it tells you why.

As devs, we see the behind-the-scenes of how sophisticated phishing and social engineering have become. Scammers are now using the same LLMs we love to generate perfect, urgent, and highly targeted messages. I got tired of being the "manual" second opinion for my friends and family, so we built ScamShield.

The "Why" Behind the Build:

Most security tools are black boxes. ScamShield is built on Explainability.

• Heuristics + LLM Analysis: We score fraud risk (0–100) and map specific triggers (payment pressure, impersonation, etc.) directly to the original text.

• The Verification Runway: We don’t just stop at "this is a scam." We provide a guided checklist to verify entities through independent sources.

• Privacy-by-Design: We aren't in the business of selling data. Scan history is stored locally in the browser (for Pro users), and we treat pasted content as ephemeral real-time analysis.

• Infrastructure: Built to scale with Redis-backed rate limiting and an optional email-inbound workflow for those who live in their inboxes.

The Stack is Ready for Testing:

Whether it’s a suspicious marketplace DM, a "urgent" IRS text, or a weird dating app message, you can now get a fast second opinion.

Try the Web Scanner (Free Tier available): 👉 [Insert Your Link Here]

I’d love for this community to kick the tires on the UX and the risk-scoring engine. If you’ve got a "highly sophisticated" scam sitting in your spam folder, feed it to the beast and let me know what you think of the red-flag mapping.

#BuildInPublic #AI #CyberSecurity #LLM #ScamShield #PrivacyFirst #InfoSec

Disclaimer: ScamShield provides automated, informational analysis only. It is not legal, financial, or law-enforcement advice. Always verify through independent, official sources before taking action.

Hey everyone, I’m Khalil. For the last few months, I’ve been obsessed with a single statistic: 90% of data breaches are caused by social engineering.

As developers, we’ve gotten really good at hardening the stack. We have automated dependency scanning, robust OAuth, and encrypted databases. But while our code is getting harder to hack, our users are getting easier to manipulate.

I’m building ScamShield because traditional antivirus is a reactive relic. It scans for "bad files," but today’s most dangerous threats are "bad conversations." With the explosion of AI-generated phishing, scammers don't need a virus—they just need to vibe their way into a user's trust using high-pressure tactics.

The Vision

ScamShield moves beyond the hardware to protect the person. It’s a sophisticated heuristics engine that identifies the linguistic markers of social engineering—urgency, fear, and authority—in real-time.

• The Goal: A "set-it-and-forget-it" guardian for families, seniors, and non-tech-savvy users.

• The Differentiator: Our competitors protect your PC; we protect you.

The Tech Stack

I’m building this in the Next.js and TypeScript ecosystem, utilizing vibe coding to rapidly iterate on the analysis engine. I’ve just finished a Chrome Extension that allows users to send suspicious text directly to our analysis HUD for a "sanity check."

What's Next?

I’m currently looking for early feedback on our "Human-Layer" approach. If you’ve ever had to help a family member recover an account because of a "Verify Identity" scam, I’d love to chat