Catch the fake SBI link before anyone clicks it.
Kavach scores newly registered domains for the tricks phishers use: swapped letters, look-alike characters, and brand names stuffed into subdomains. It runs fully offline, with no WHOIS lookup.
Got this link in a message? Don't open it.
Use this result somewhere else
A fake bank page only needs one believable link.
New domains go live every hour. A few of them borrow a bank or app name, sit unnoticed until an SMS campaign starts, and disappear in days. Kavach looks at the name itself, so it can raise a flag on day zero.
Got a suspicious SMS? Paste the whole thing.
Kavach pulls every link out of the message, scores each domain, and marks the pressure phrases scammers lean on, like account blocks, KYC deadlines and OTP requests.
Check a whole file of messages
Upload a CSV with a message column (and an optional sender column), or a text file with one message per line.
This needs the server. Run python server.py in the project folder, then reload this page.
Only have a picture of it? Drop in the screenshot.
Kavach reads a screenshot of a scam SMS, a WhatsApp chat or a website, then runs the same checks as the rest of this page. It reads the address and the page text, and it never opens any link it finds.
Your image never leaves this computer. It is read in memory and is not saved.
Add a screenshot on the left. The text it reads, the links it finds and the verdict appear here.
Scan a whole day of new registrations at once.
Upload a .txt or .csv list, paste domains, or drop a file on this section. Kavach scores every domain on your device and puts the riskiest at the top.
| Domain | Closeness | Risk | Verdict | Top reason |
|---|
Does this page copy the real one?
Upload saved copies of a suspicious page and the genuine page. Saved HTML gives a content similarity, and screenshots give a visual similarity. Nothing is fetched from the internet, and the files are only read, never opened or run. This is extra evidence shown next to the verdict: it does not change the risk score.
This needs the server. Run python server.py in the project folder, then reload this page.
Suspicious page
Genuine page
See the fake domains a scammer could register for your brand.
Pick a brand or add your own. Kavach generates the look-alike names attackers try first and shows how many its engine would flag. Use the list to watch new registrations, or register the riskiest names before someone else does.
Three passes over the name. That's the whole trick.
Every domain goes through the same pipeline, in this order. Nothing leaves the machine.
Clean the input
Strip the protocol, path and www. Decode punycode so xn-- labels show their real characters. Split off the suffix, including Indian ones like .co.in and .gov.in.
Score the name
Rules compare the name with the watched brands: fake letters, one-typo edits, brands inside longer names or subdomains, risky endings, scam words.
Show why
Every verdict shows the signals behind it, and a list can be scored at once.
The page is using its built-in browser engine right now: rules only, with different cut-offs (70 and 40). Run python server.py in the HACKFEST folder to switch to the full engine with the ML model.
What the engine looks for
Brand impersonation carries the most weight. The structural signals push a borderline domain over the line.
In a message
No WHOIS. No API key. Still useful.
Registration data is often hidden behind privacy services, rate-limited, or slow to appear for a new domain. Kavach works from the name alone, so it gives an answer instantly and keeps the checked links private.
Found a fake site? Report it.
Reporting gets fake domains taken down and warns the next person. These are the official channels in India. On any risky result, Copy report gives you Kavach's findings ready to paste.
National Cyber Crime Helpline
Call if money left your account or you shared bank, card or UPI details. Tell your bank as well.
National Cyber Crime Reporting Portal
Report the fraud online. Keep screenshots of the message, the link and any payment.
Chakshu on Sanchar Saathi
Report suspicious calls, SMS or WhatsApp messages to the Department of Telecommunications, even if you didn't lose money.
CERT-In
Email India's national cyber security agency about the website, and paste the report Kavach copies for you.
Kavach is an independent tool and is not affiliated with these services.
The next fake bank link is being registered right now.
Paste any domain above and see how Kavach reads it, signal by signal.