Sentinel-7 Anti-Cheat

The server sees everything.

Roblox runs your character on your machine, so Sentinel-7 trusts nothing it tells us. It re-simulates physics, recomputes every hit, validates every remote, and reads memory + honeypot traps, judging the result, not the claim.

Signals become a verdict.

A scoring engine stacks weak signals and fires instantly on confirmed ones. Lag, teleporters and real mechanics get grace, so clean players are never touched while exploiters spike fast.

Erased. Everywhere.

Confirmed exploiting ends the session and triggers a permanent ban: account, known alternate accounts, and the device, across every Lucky 7 game. Every action is logged with evidence.

Hovering, gliding and ascending without a jump, judged from real airtime, not what the client claims.

Movement faster than physically possible, and warps beyond any legitimate distance.

Phasing through walls is caught by raycasting the path between every position the server sees.

Runaway velocity and any attempt to throw another player: the server never lets you move someone else.

The engine behind fly / fling / noclip: illegal forces injected into a character. One of our hardest signals.

Health is server-owned. Refusing to take damage or sitting above max HP is confirmed and erased.

Stretched limbs, shrunk hitboxes, edited humanoids and tampered stats are reset and flagged.

Every hit is recomputed on the server: out-of-range hits and inhuman aim are rejected on sight.

Injected cheat UIs leave a memory + instance + honeypot fingerprint the sentinel hunts for.

Every gameplay request is type-checked, rate-limited and re-validated: the server is the only authority.

Economy and inventory changes are transactional; impossible balances are rolled back instantly.

Trying to disable, hook or spoof the sentinel itself is, by design, its own kill signal.

Real-time filtering already flags slurs, scams and harassment for review; the roadmap takes it further: automatic mutes for confirmed violations and direct escalation to human moderators for the gray-area cases.

As voice chat spreads across our games, Sentinel-7 is being extended to flag harassment and abuse signals in voice the same way it already does for text.

Beyond matching banned accounts and devices, we are building pattern-based correlation so an exploiter’s play style can be recognized on a fresh alt before it does any damage.

Today’s detections run on tuned heuristics and hard thresholds. We are training a model on shadow-mode data to catch subtler macro and bot behavior the current rules miss.

Community members with a track record of accurate reports will get their flags routed for expedited review, cutting the time between a report and a confirmed action.

Periodic published stats on detections and bans issued, so players and studios can see the system is actually catching cheaters, not just claiming to.