01 · TEXT
Message analysis
SMS, WhatsApp, email. Every phrase that applies pressure or asks for something sensitive is pulled out as evidence.
SCAMSHIELD / THREAT ANALYSIS
Don’t get scammed.Know what you’re looking at.
Paste a suspicious message or link. ScamShield breaks down the threat before you act — the score, the evidence behind it, and what to do next.
- LOCAL RULE ENGINE
- AI ANALYSIS
- LINKS NEVER OPENED
Have a screenshot? Use the full analyzerPNG, JPEG or WebP · up to 3 MB
LIVE ANALYSISTHREAT SCANNER
READY0/8,000
SHIFT + ENTER FOR NEW LINE
- INPUT RECEIVEDWAITING
- RULE ENGINEWAITING
- URL STRUCTUREWAITING
- AI ANALYSISWAITING
- RISK ASSESSMENTWAITING
NO THREAT REPORT YETSubmit an input to begin analysis.
SCAMMERS PRETEND TO BE
- Your bank
- Delivery firms
- Employers
- Payment apps
- Government offices
THREE WAYS IN
Whatever reached you, check it. Scams don’t look like scams anymore.
The obvious tells are gone. Modern scams copy the exact tone and branding of organisations you already trust. So ScamShield looks past the surface to what the message is actually asking you to do.
02 · URL
Link analysis
Links are taken apart as text — never opened — to expose lookalike domains, buried subdomains and login-shaped paths.
03 · IMAGE
Screenshot analysis
Upload a screenshot of a chat or email. The text is read from the image, then scored by the same engine as a pasted message.
PNG, JPEG or WebP · up to 3 MB
TRY IT HERE
Watch the engine think.
Edit the message and see ScamShield’s pattern checks respond as you type. These are the real rules the server runs first — running right here in your browser.
INPUT · EDITABLEFICTIONAL SPECIMENS
Runs locally · nothing sent109 chars
PATTERN ENGINE · LIVE4 SIGNALS
66/ 100HIGH RISK
- Credential requestHIGH
- Urgency pressureHIGH
- Account threatHIGH
- Possible brand impersonationHIGH
This is the deterministic layer only. The full analysis adds AI reasoning, an attack path and specific next steps.
HOW IT WORKS
Four steps. About ten seconds.
You end up with more than a verdict — you end up understanding the message, which is what helps the next time one arrives.
- 01
Paste
Drop in a suspicious message, a link, or a screenshot. Nothing is stored unless you choose to save it.
- 02
Analyze
A deterministic security engine checks the structure, then AI reads the intent behind the words.
- 03
Understand
A risk score, the exact phrases that triggered it, and the attack path they lead to.
- 04
Act
Clear, specific next steps for that kind of scam — not a generic warning to be careful.
UNDER THE HOOD
We don’t just ask an AI if it’s a scam.
An AI asked “is this a scam?” will answer confidently either way, and you have no way to check it. So the deterministic engine runs first and runs always. If the AI layer is unavailable, ScamShield still produces a complete report from pattern analysis alone — and tells you so.
01
User input
Message, URL or screenshot. Validated and size-limited before anything runs.
02
Deterministic security engine
Pattern rules and URL structure analysis. No AI involved, and it runs on every request.
03
AI analysis
Gemini reads intent and pressure. Submitted content is fenced as untrusted data, never obeyed as instructions.
04
Risk engine
Both scores are deduplicated, weighted and combined into one auditable number.
05
Structured threat report
Score, evidence, attack path and contextual actions — every claim traceable to a detected signal.
WHAT SCAMSHIELD DETECTS
The mechanics behind almost every scam.
Scams vary enormously in story and almost not at all in structure. These four mechanics carry most of them.
Urgent pressure
Deadlines, threats and countdowns exist to stop you checking with anyone else. A real organisation will let you call them back.
"Your account will be blocked today"
Brand impersonation
We compare the brand a message claims against the domain it actually links to, and flag the gap between them.
Says SBI → links to sbi-secure-login.example
Suspicious links
URLs are parsed as text: raw IP hosts, punycode, buried subdomains, shorteners, login-shaped paths. Never opened.
verify.account.secure.xyz/login
Credential theft
Requests for passwords, OTPs, card details or a UPI PIN. An incoming payment never needs your PIN.
"Enter your UPI PIN to receive ₹5,000"
Before you click, check.
Turn a suspicious message into a clear answer in seconds.