Server-authority rewrites
High-risk events are moved behind server checks so impossible state never lands.
DEMONTIME · GUARD
Server-authority anticheat — no client trust, explainable, reversible.
Cheating,
caught server-side.
DemonTime Guard is the security product behind Demon Scripts: server-authority rewrites, exploit rejection, live telemetry, AI anomaly review, and protected service-side logic. The demo is open now while the full runtime ships toward launch.
Watch it catch a cheat
DEMONTIME // OPSread-only · logged · reversible
00 / PRODUCT FILM
Reject first. Review smarter.
A Guard ad focused on the product in action: a bad client claim hits the server boundary, fails validation, then becomes a reviewable evidence packet instead of blind enforcement theater.
server truthevent rejectionAI evidencereversible review
00 / PRODUCT · DEMONTIME GUARD
Not another detector. A server-truth layer.
Most FiveM anticheats market detections, panels, spectate, maps, cloud bans, and quick setup. Guard is built to protect the state cheats actually want to change: money, items, jobs, weapons, permissions, and high-value server logic.
Reject the lie.
Explain the signal.
The strongest anti-exploit move is refusing to believe client claims. Guard validates the action against server truth first, then uses AI for patterns that rules alone miss.
AI on owned telemetry
Guard reads server-side event history and produces reviewable anomaly evidence instead of blind rage-bans.
Protected service logic
Sensitive anticheat, license, and live-AI logic stays behind an encrypted service boundary where dumping a client does not reveal the brain.
01 / DEMO · TIER-A
Injected money, rejected.
A mod menu tries to give itself a million dollars. Watch Guard reject it before it ever lands — because the client never had the authority.
guard · server-authority check
tier-a · auto-act
Tier-A: server-observed, auto-acted. The client never had the authority — the money had no legitimate source event, so it was rejected before it touched the balance.
02 / DEMO · TIER-ML
Economy fraud, flagged by AI.
Not everything trips a hard rule. Subtle fraud shows up as a pattern — so we run a model on our own telemetry, explain it with SHAP, and place a reversible hold.
guard · anomaly model
isolation forest · cpu-only
Server-side ML on real telemetry. Explainable (SHAP), reversible, no kernel driver — the model reads our own dtrp_datalogger events, not the player's machine.
03 / MARKET COMPARISON
Different fight. Different layer.
This is not a dunk contest. ChocoHax/Lyxia and Reaper show where the market already is. Guard is aimed at the server-authority and evidence gap that normal detection stacks do not fully own.
Panel-first protection
Public positioning emphasizes real-time protection, a remote panel, live map, multi-spectate, admin menu, backdoor/resource scanning, detection methods, and cloud bans.
market baselineLua anticheat stack
Public positioning emphasizes FiveM-focused detection, client/server checks, automated enforcement, analytics, custom rules, a live web panel, and quick install.
market baselineServer-truth security layer
Guard starts where exploits pay off: impossible server state. It rejects invalid money/items before they land, flags subtle economy fraud, and explains every reviewable action.
our wedgeCompetitor notes are based on public product positioning, not an internal bypass audit. Guard is in prelaunch demo access; serious servers should run layered security until the full runtime ships.
04 / HOW GUARD IS BUILT
Server-brain first. Three tiers.
No client trust anywhere. The server is the only source of truth — each tier earns more autonomy the more certain the signal is.
Tier-A
Server-observed → auto-act
Events the server itself witnessed. If a value can't be real — money with no source, an item that never existed — it's rejected and reversed automatically.
auto-actTier-B
Client probe → score only
Optional client-side probes can raise suspicion, but they never decide. A dumped or spoofed client only changes a score — it can't authorize anything.
score onlyTier-C / ML
Signatures + AI → flag
Signatures, OCR, and a server-side anomaly model surface what looks wrong. These flag for review with a reversible hold — never an instant rage-ban.
flag for review
server-authority
explainable (SHAP)
reversible enforcement
hard-capped
AI on our own telemetry
05 / WHY IT'S HARD TO BEAT
Server-authority is the moat.
You can't cheat a system that never believed you in the first place.
Nothing the client says is taken on faith
Every client claim is validated against server truth. A "give me money" packet isn't a request Guard grants — it's a claim Guard checks, and rejects when no real event backs it.
Reverse-engineering the client buys you nothing
The decisions live server-side. A dumped, patched, or spoofed client can only move a Tier-B score — it can never authorize an action the server didn't observe.
The AI learns from our own live telemetry
The anomaly model trains on real dtrp_datalogger events from a running server — patterns that are hard to fake because they're measured downstream of every action.
It won't rage-ban your real players
Enforcement is hard-capped and reversible, ML decisions are explained with SHAP, and uncertain signals flag for review instead of acting. Catch cheats, keep your community.
06 / WHAT OWNERS CAN BUY
Start with an audit. Graduate into Guard.
Guard is not a one-file miracle. The product is a security path: harden the server, move high-value logic server-side, then add the runtime and AI layer when your stack is ready.
Stark Shield audit
From $299
Exploit audit, event risk map, resource review, and a prioritized hardening report.
- Client-trust findings
- Money/item exploit checks
- Server-authority roadmap
Guard runtime
Waitlist
Runtime rules, event wrappers, anomaly evidence, reversible holds, and owner-facing review flow.
- Tier-A rejection
- Tier-ML anomaly review
- Explainable evidence trail
Protected boundary
Scoped
Encrypted service boundary for sensitive logic, licensing, anticheat scoring, live-AI actions, and continuity terms.
- Escrow-ready installs
- Protected scoring logic
- Audit logs and access controls
Want this on your server?
Demo access is open during prelaunch. Start with a Stark Shield audit, join the Guard waitlist, or commission the server-authority rewrites your stack needs first.