Fake-browser detection is live on every plan, including the free one. See how it works →
[ 200 OK ][ /v1/score ]
New: catching bots that fake their browser →

Stop fraud before it costs you money

Fraudsters rent real home internet connections to hide where they are, and run browsers built to fake who they are. Kaidn reads the handshake underneath, the part they have to send before your page even loads.

No credit card. One API call. No sales team to get past.

device intelligence — awaiting consent

This reads how your browser renders graphics, audio and text to build a device fingerprint — exactly what Kaidn does for customers. We don’t run it on you without asking.

What it reads →

Built for the operators big fraud vendors ignore. One API for rewards, affiliates, crypto, SaaS, games or fintech. No enterprise contracts, no sales calls.

Rewards
Affiliates
Crypto
SaaS
Games
Fintech
THE USUAL CHECKS

Both sides can be bought

  • Clean home IP addresses cost pennies. Fraudsters buy them by the thousand, so checking the address stopped working years ago.
  • The browser can be faked just as cheaply. A whole industry sells tools that give every fake account a convincing, different-looking device.
  • So you turn away real customers while the professionals walk straight through.
WHAT WE ADD

The layer underneath

  • We read the handshake, not the claim. How a connection is opened is set by the software, not by anything the page can edit.
  • Then we check it against the whole network. Caught somewhere else this morning means known here, even on a new address and a new device.
  • You get allow, review or block with the evidence in a sentence, not a score you have to take on faith.

How that actually works →

01 / What it catches

Never a number you can’t explain.

Every answer tells you what we found, how much it counted, and why, in a sentence your support team could send to the customer. No mystery score.

FAKE BROWSERSIndustry first

The handshake gives it away

A bot can change everything its browser claims to be. It cannot easily change how it opens the connection in the first place, and we read that before your page even loads.

handshake
"t13d3112h2_8daaf61527df_845e22731872"
SHARED INTELLIGENCE

One attack teaches the network

When one site on the network catches a fraudster, everyone else already knows them. Nothing readable is shared, so nobody, us included, can see another operator's customer list.

median lookup6 ms
what we sharehashes, never emails
tierGrowth and up

Where they connect from

VPNs, proxies, and the rented home connections fraud rings hide behind.

IP & ASN intelligence

The same device, again

Recognise a returning browser after they clear cookies and change their details.

Device identity

Too much, too fast

Ten signups from one place in a minute are not ten customers.

Velocity

One inbox, many accounts

Throwaway addresses, and the tricks that turn one mailbox into fifty.

Email identity
02 / Free tools

Run a check without an account

Rate limited, no key required. Results appear inline.

03 / User lifecycle

Fraud caught at every step

01 / SIGNUP

Block the farm

Bulk account creation, fake inboxes and residential proxy farms.

02 / LOGIN

Catch the takeover

Account takeover, credential stuffing and session hijacking.

03 / CHECKOUT

Stop card testing

Promo abuse and chargeback-prone proxy orders.

04 / WITHDRAWAL

Guard the money

Device change or a suspicious geo shift before funds leave.

05 / CUSTOM EVENT

Anything else in your stack

Referral links, API keys, support requests. Name the event, set the rules, POST it.

See the event list
FIELD GUIDES

How to stop it, by fraud type

How each attack actually runs, the defences that look sensible and fail anyway, and the signals that catch it.

All fraud types
COMPARISONS

Honest comparisons, including where we lose

What each tool is genuinely built for, and when you should buy theirs instead of ours.

All comparisons
04 / The honest limit

Where it falls down

And the part we are honest about

None of this catches everything, and we publish where it fails rather than waiting for you to find out. A browser built to spoof the operating system it is already running on defeats our engine and the market leader’s alike. We measured it, wrote it up with the numbers, and shipped the post. Read the measurement →

[ SDK ]
05 / Integrate

Integrate in minutes

IF YOU ARE BUILDING IT

One POST, one answer

A REST call or the npm package, whichever you prefer. Send the fields you already collect; anything you do not have is optional. No sales call, no onboarding project, no contract to sign before you can test it on real traffic.

IF YOU ARE REVIEWING IT

See why, and change your mind

The dashboard shows every decision with the signals that fired and what each one counted for. Disagree with an outcome and you can move that signal’s weight yourself, or the threshold, without asking us.

Use the @kaidn/sdk npm package, a plain REST call, or the browser tag. Score one event or a million through the same endpoint. Every field except event is optional.

score.ts
// npm install @kaidn/sdk
import { Kaidn } from "@kaidn/sdk";

const kaidn = new Kaidn({ apiKey: process.env.KAIDN_API_KEY });

const { verdict, reasons, reason_text } = await kaidn.score({
  event: "cashout",
  user_id: "u_8231",
  ip: req.ip,
  email: user.email,
  device_id: fingerprint,            // from @kaidn/fp in the browser
  event_country: offer.country,      // optional geo expectation
});

if (verdict === "block") return deny(reason_text);
if (verdict === "review") await queueForReview(user, reasons);

// the SDK covers every endpoint:
await kaidn.check.ip("3.5.140.1");
await kaidn.batch.score(rows);       // bulk, up to 1000 rows
await kaidn.label({ label: "chargeback", event_id });
06 / MCP server

Investigate fraud by asking

Connect Claude, Cursor or any MCP client to your tenant. The agent gets the evidence behind every verdict, not a bare score. Big vendors gate this to invited enterprise accounts.

01 / ADD IT
claude mcp add kaidn \
  --env KAIDN_API_KEY=your_key \
  -- npx -y @kaidn/mcp

Read-only by default. The two tools that change anything need an explicit --allow-writes, and a per-session quota ceiling means an agent stuck in a loop cannot burn your month.

02 / ASK IT
Why was this signup blocked?
Show everything that touched this device.
Are these two accounts the same person?
What is in review, worst first?
07 / Questions

Billing questions, answered here

Is there really a free tier?

Yes: 10,000 events a month, forever, no card. It includes all engine checks and verdicts so you can evaluate Kaidn on your own traffic. AI-written reasons unlock on the paid plans.

How am I billed, and what about tax?

Paid plans are billed in advance, monthly or annually, and renew until you cancel. Payments run through Paddle, our Merchant of Record, which adds any applicable VAT or sales tax at checkout and sends your receipt. We never see your full card details.

What counts as an event?

One call to POST /v1/score is one event. The free-tool checkers and dashboard views do not count against your allowance.

What happens if I go over the limit?

We will let you know before it bites. Depending on your plan we may throttle, queue overage for billing at the published rate, or ask you to upgrade. We will not silently stop scoring your traffic.

Can I cancel or get a refund?

Cancel any time from your dashboard and you keep access to the end of the period you have paid for. Your first paid subscription is covered by a 14-day money-back window.

Do you offer annual or custom pricing?

Annual billing is available on every paid plan and saves two months. Enterprise is fully custom: email support@kaidn.io and we will size it with you.

Ready to stop paying fraudsters?

Start free with 10,000 events every month. No credit card. Your first integration takes less than five minutes.