How the score worksEvery number, and where it comes from.

The score runs from 0 to 100; higher means fewer risk signals. It is a weighted average of the factors we could measure for that domain at that moment, and nothing else.

The factors and their weights

FactorSourceWeight
Domain age

Registration date from the registry's RDAP record.

rdap.org → the registry45%
HTTPS certificate

A TLS handshake on port 443; the certificate must chain to a trusted authority and match the name.

Direct connection30%
Email authentication

SPF record on the domain and DMARC record at _dmarc.

DNS TXT records25%

Domain age

Age is counted in days from the registration date to the moment of the check.

Under 1 month
5/100
1 to 3 months
15/100
3 to 6 months
35/100
6 to 12 months
55/100
1 to 2 years
70/100
2 to 5 years
85/100
5 years or more
100/100

HTTPS certificate

Valid for the name
100/100
Offered but invalid
10/100
Connection refused
0/100
Timed out or unreachable
not checked

Email authentication

SPF and DMARC
100/100
One of the two
60/100
Neither
20/100
DNS did not answer
not checked

Crypto signals

Run on every check. Each is one-sided: it enters the average with the weight below only when it finds something, and "no match" is left out, because not being on a list or not looking like a brand says little. A phishing-list entry or a seed-phrase form also caps the score.

SignalSourceWhen found
On a crypto phishing list

The name or a parent of it is on one of the lists below. We show the list and the date: the list's own date when it gives one, else our first fetch that contained it.

Public lists, fetched dailyweight 100, score 0, capped at 10
Seed phrase or drainer in the page

An input or textarea named or labelled for a seed phrase, recovery phrase, mnemonic or private key; a form telling you to enter one; or the name of a known drainer kit (Inferno, Angel, Pink, Monkey, Venom, Ace, Medusa, Nova, Atomic and other drainer kits, or drainer.js). The matched marker is named.

Homepage HTMLweight 60, score 0, capped at 25
Lookalike of a crypto brand

Compared with the official domains of the largest exchanges, wallets, coins and protocols: international letters that read as the brand (punycode), swapped characters (0 for o, 1 or i for l, rn for m, vv for w), the brand plus bait words, or a name one letter off (two for long names). Strong: built like a phishing name. Weak (largest brands only): a plain typo, the brand plus any word, or the brand on a free hosting platform.

CoinGecko and DefiLlama official sitesstrong: weight 40, score 5; weak: weight 25, score 40
New domain with bait words

Registered less than 90 days ago by the RDAP record, and the name contains a bait word. Unknown age means not checked.

RDAP and the nameweight 30, score 0
Listed as an official site

CoinGecko or DefiLlama lists this site as the website of an exchange, coin or protocol. A positive signal; we name the entry. It confirms the address, not the quality of the service.

CoinGecko and DefiLlamaweight 40, score 100

Bait words: airdrop, airdrops, claim, claims, reward, rewards, bonus, giveaway, gift, free, double, support, help, helpdesk, service, care, desk, ticket, live, chat, wallet, wallets, connect, sync, verify, verification, validate, validation, rectify, restore, recover, recovery, unlock, unlocked, secure, security, safe, protect, login, log, signin, sign, auth, account, accounts, kyc, update, upgrade, migrate, migration, refund, refunds, redeem, mint, drop, dapp, dapps, official, web, app, portal, event, promo, password, reset, logon, access, authorize, approve, vote, voting, allocation, eligibility, eligible, checker, distribution, points, season, unstake, withdraw, withdrawal, 2fa.

The phishing lists

A name the same list allow-lists is never reported. Removals are up to each list's maintainers.

The official sites

  • CoinGecko public API. Free public API; attribution to CoinGecko required on the free plan (shown on every page that uses it).
  • DefiLlama public API. Free and open API, no key; attributed to DefiLlama on every page that uses it.

We are not affiliated with any list or aggregator.

When a factor cannot be checked

A lookup can fail: the registry does not answer, the server times out, DNS returns an error. That factor is then shown as not checked and is removed from the average; the remaining weights are scaled up. It never counts as a fixed or guessed value. With fewer than 2 measured factors there is no score at all, only the findings.

The result page always says how many factors were checked, so a score built on two factors is never mistaken for one built on three.

Bands

80 to 100
Few risk signals
60 to 79
Some risk signals
40 to 59
Several risk signals
0 to 39
Many risk signals

The bands describe signals, not intent. We never label a site a scam or safe.

Deliberately not scored

  • General blocklists. The reliable malware and phishing lists for all sites need an API key or forbid this use, so we link to Google Safe Browsing and VirusTotal. Only the openly licensed crypto phishing lists below are checked.
  • Traffic rank. No free source is wired in, and a guessed rank would be an invented figure.
  • Registrar and owner privacy. Shown for context. Hiding owner details is normal and required by law in many countries, so it is not a risk signal.
  • Reviews. We do not collect opinions. The homepage HTML is read once, only for the crypto markers below.

Freshness and limits

A result is kept for 24 hours and then checked again on the next visit. Checks are rate-limited per connection. We only connect to public addresses on port 443. One GET request reads the homepage HTML (at most 1 MB, redirects only to https on public addresses); nothing on the page is run, no script or image is fetched, and a check never logs in or submits a form.