Solutions

One API. The right signal for each problem.

"Ban the device" means something different in a game, a FiveM server, a SaaS login, and a licensed app. HWID-Guard weighs the signal that actually fits — not one blunt fingerprint for everything.

Why this matters

A generic HWID check bans innocent players and misses real abusers. The difference between good and useless is weighting the right identifiers for the case. That's the product.

Indie games & launchers
Signal · weighted hardware composite

The problem

A player is banned for cheating and reappears in minutes on a new account — same PC. A small team can't run an anti-cheat department, and needs a signal that works in the game and the launcher, before login.

How HWID-Guard handles it

A weighted composite of hardware identifiers, scored server-side, with a threshold that needs multiple signals — so swapping one part won't ban an innocent player. VM detection stops players spinning up a virtual machine to reset their hardware.

Composite weights
smbios_uuid55
baseboard_serial45
cpu_id40
disk_serial20
No single signal reaches the threshold alone — the composite decides. Optional TPM anchor for high-value bans.
FiveM & RP communities
Signal · native hardware token

The problem

A cheater kicked from an RP server is back in minutes on another Rockstar account — the account is disposable, the PC isn't. Small communities rely on manual reports; by the time staff re-ban, the damage is done. Multi-server networks want a ban to optionally carry across servers.

How HWID-Guard handles it

FiveM exposes a strong, anti-tamper hardware token (GetPlayerToken) that most platforms don't have. Account and network identifiers (steam:, ip:) are ignored — they're not the device. Drop-in Lua SDK.

Identifiers
fivem_token80
fivem_license25
Here one strong signal is enough: a BAN on a fivem_token match alone — the hardware token doesn't disappear when they switch accounts.
SaaS & anti-abuse
Signal · origin classification + evidence

The problem

A free tier gets hammered with throwaway accounts — trial abuse, coupon farming, credential stuffing, bots clearing signup at inhuman speed. IP bans miss residential proxies; CAPTCHA alone adds friction and doesn't stop re-entry: the same automated machine comes back on a new account.

How HWID-Guard handles it

The browser has no hardware HWID — so the strong signal here is origin classification. The client reports raw signals (webdriver, headless, automation) and the server decides bot vs person; a validated CAPTCHA/liveness pushes toward person. VM detection catches datacenter bots.

Verdict inputs
source_kindbot / person
vm_detectedflag
captcha / livenessevidence
A failed check doesn't ban outright — it asks for REVIEW / step-up. No hard block on a legitimate user.
Licensed software
Signal · strict hardware composite

The problem

Paid software gets cracked, or one license key is shared across many machines. Blocking by the key alone doesn't work — a string is trivial to copy. What has to stay bound is the device, and you need visibility into how many devices actually activated a license.

How HWID-Guard handles it

The desktop composite applies, but with more rigor — a false positive here blocks a paying customer, so the threshold and the review path matter more than in a free game. Bind activations to hardware, count real devices per license.

Composite (strict)
smbios_uuid55
baseboard_serial45
bios_serial30
Higher rigor: a low-assurance match goes to REVIEW, not a hard block — you don't lock out a real customer over one changed part.
OTServer & private MMO servers
Signal · native desktop composite

The problem

A botter or multi-client abuser is banned and reappears on a fresh character, same machine — and the level/item economy rots for everyone. Owners are usually solo, with no anti-cheat team; IP bans punish shared connections and account bans cost the abuser nothing.

How HWID-Guard handles it

A native collector in the client (OTClient C++) reads the desktop hardware composite, weighed server-side; IP and account are ignored — they're not the device. VM detection stops the "fresh virtual machine" reset trick. Built by someone who runs one of these servers.

Composite weights
smbios_uuid55
baseboard_serial45
disk_serial20
Same-device match on a new character → BAN at login. Swapped one part? The composite still needs a threshold — no accidental ban.
Competitive & ranked
Signal · device-to-accounts correlation

The problem

One human spins up many accounts to smurf in ranked, dodge MMR, or evade a competitive ban — wrecking matchmaking for everyone else. The accounts are free and infinite; the machine behind them is not.

How HWID-Guard handles it

Device reputation correlates many accounts to one device — you see the single machine behind N accounts and can cap accounts-per-device or flag the smurf. Ambiguous cases (a shared family PC, siblings on one machine) go to REVIEW, not an automatic ban.

Signal
device → accountscount
hardware compositematch
vm_detectedflag
Many accounts, one device = a smurf signal — but a shared machine is real too, so ambiguity is REVIEW, not a blind ban.
Marketplaces & communities
Signal · device banlist

The problem

A scammer, a banned seller, or a repeat-offender returns under a new profile and email to keep abusing buyers or the community. Moderation bans the account; the person and their machine are unchanged, so they're back the same day.

How HWID-Guard handles it

Ban the device, not the profile — the banlist keeps that hardware out across every new signup. A public or shared machine (library, internet café) goes to REVIEW so you don't wall off a whole location over one bad actor.

Verdict inputs
device banlistmatch
hardware compositescore
Known-bad device on a new profile → BAN. Shared/public device → REVIEW, not a blanket block.
Bookings & reservations
Signal · bot classification + device correlation

The problem

Scalpers and bots hoard scarce slots — restaurant tables, appointments, event tickets, exam/visa bookings — across dozens of throwaway accounts, then resell or no-show. Banning the account or the email does nothing: it's one machine behind many identities, and a repeat no-show abuser is back the same day with a fresh signup.

How HWID-Guard handles it

Booking is mostly web, so the strong signals are origin classification (webdriver/headless/automation → bot) and device-to-accounts correlation — N accounts booking from one device is the scalper pattern. A banned no-show device stays out across new signups.

Verdict inputs
source_kindbot / person
device → accountscount
device banlistmatch
Automation hoarding slots → BAN before the slot is taken. A family booking four seats from one laptop is normal — ambiguity goes to REVIEW, never a blind block.

Your case isn't listed? The signal still fits.

These are the patterns we've mapped — games, RP servers, SaaS abuse, licensed software, ranked smurfing, marketplaces, bookings, and fraud/anti-abuse stacks. The same API adapts to whatever your device story is. Tell us your setup and we'll point you at the right weighting.

Get your API key