Paste it before
you trust it.
A link, a message, an email, a payment request. Electryo tells you in plain words what it actually is, why, and the one thing to do next.
Why it works
No single detail gives a scam away. All of them together do.
Every part of a convincing message is defensible on its own. Electryo is built to look at them at the same time, which is the only place the pattern shows up.
01
It reads the whole trick
A scam is rarely one bad detail. It is a plausible name, a fresh domain, a deadline and a login box, each forgivable alone. Electryo weighs them together, which is where the pattern becomes obvious.
02
Evidence decides, not tone
A confident message is not a safe one. The score is arithmetic over what was actually found, computed before any model writes a sentence about it.
03
One answer, one next step
Three verdicts, a plain-language reason, and one thing to do. No dashboard, no severity matrix, nothing to interpret.
What it checks
46 kinds of evidence, across 8 categories.
Each one has a stable code and a fixed weight, so a verdict can always be explained by pointing at the specific thing behind it.
Reputation
Known blocklists
Whether the address is already recorded as phishing, malware or unwanted software.
The address itself
Brand lookalikes
A well-known name sitting in a domain that company does not own - apple-id-verify.something-else.tk.
Disguised characters
A Cyrillic а inside аpple.com reads as English to you and as a different website to your browser.
How old the website is
A bank whose website was created nine days ago is not a bank. Age is weighted in weeks, months and years.
Domain endings and nesting
Endings with a disproportionate share of abuse, bare IP addresses, and subdomains stacked to bury the real host.
The connection
Certificate details
Expired, self-signed, issued days ago, or issued to a completely different site than the one you are on.
Unencrypted pages
A login page served over plain http, where anything typed into it travels in the clear.
Where it goes
The real destination
Shorteners and redirect chains are followed to the end before anything is judged, on our infrastructure rather than yours.
Link text that lies
Text reading paypal.com wrapped around a link that goes somewhere else entirely.
The page
What it asks for
A password field, a card form, a seed phrase - and whether the site asking has existed for more than a fortnight.
Copied identity
Layout and wording lifted from a real company, hosted somewhere that company has nothing to do with.
The sender
Who really sent it
Email headers behind a friendly display name, plus SPF, DKIM and DMARC results.
Where replies would go
A Reply-To pointing at a mailbox that has nothing to do with the company in the signature.
The wording
Pressure and threats
Deadlines, suspended accounts, consequences for hesitating. Urgency is a signal, not a formality.
What it wants from you
Passwords, card details, gift cards, crypto transfers, bank transfers, prizes you never entered for.
Claimed authority
A message presenting itself as your bank, the tax office, the police or a courier.
Attempts on us
Instructions aimed at the tooling
Text written to talk an automated checker into clearing it. Finding this raises the score; it can never lower it.
46
named findings, each with a fixed weight
8
categories of evidence behind a verdict
6
investigation tools, every one read-only
3
possible answers, and no maybe
Questions
The ones worth asking first.
Do I have to click the link to check it?
No, and you should not. Copy the address as text and paste it. Everything is fetched from our own infrastructure, so the suspicious page never meets your browser or your logged-in session.
What can I paste?
A link, a whole text message, a full email including its headers, a payment request, a job offer, a marketplace listing. If there is no link in it, the wording, the sender and the payment details are still checked.
Does a safe verdict mean it is definitely safe?
It means that of 46 things we check for, none turned up. That is a strong second opinion rather than a guarantee. Anything asking for a password, a payment or remote access deserves the same caution whatever the verdict says.
Can a message talk Electryo into clearing it?
It can influence the sentence you read. It cannot influence the answer. The score is computed from weighted evidence before any model writes prose, some findings force a minimum threat level, and an explanation that contradicts the findings is thrown away and rebuilt from them.
How long does a check take?
Usually seconds. Every investigation runs inside a fixed budget of tool calls and seconds, so it always finishes rather than hanging on one slow provider.
What happens if one of the checks fails?
It is reported. A failed provider is recorded as a finding of its own and lowers the stated confidence, so you can tell the difference between 'we looked and it was fine' and 'we could not look'.
What does confidence mean, exactly?
How much evidence the verdict rests on - how many different kinds of check contributed, and how many could not be completed. It is not a measure of how dangerous something is. A dangerous verdict at low confidence is still a dangerous verdict.
Do you keep what I paste?
Yes, so your history works and so you can return to a verdict later. It is visible only to the account that submitted it.
Is this an antivirus or a browser extension?
Neither, and it does not replace either. It replaces the moment you would otherwise ask someone 'does this look real to you?' and wait.

