RoSentry vs UltimateAntiCheat

RoSentry is a cloud-based, server-side anti-cheat for Roblox games. It learns typical behavior for each player and flags differences with evidence, comparing observations such as movement speed, weapon range, and player stats with that player’s baseline. A new installation begins in learning mode: it records findings as shadow data without automatically punishing players. Developers can add custom checks in Luau alongside built-in detections. Moderation tools cover investigations, audit logs, ban waves, and appeals; Discord integration lets moderators look up or ban players, work appeal threads, and receive alerts. Installation involves placing RoSentry.rbxm in ServerScriptService and setting a Roblox secret for api.rosentry.com. The maker says the in-game SDK does not capture screens, analyze aim or mouse movement, or use hardware IDs. If its backend is briefly unavailable, the join check retries instead of ejecting players. The free plan supports one game and up to 100 concurrent players; paid plans start at $19 a month.

UltimateAntiCheat is an open-source usermode anti-cheat project for Windows, described by its maker as an educational reference rather than commercial software. Its listed checks cover memory integrity, manually mapped and unsigned modules, debugger activity, process creation, and handles opened to a protected process. Prevention options include remapping protected sections, limiting certain new threads, hiding threads from debuggers, and applying mitigation policies. A client-server option can send a heartbeat to clients every minute; networking can be switched on or off in settings. The maker says the project avoids file-system scanning and instead checks byte patterns in processes that are already open or created while it runs. The README recommends integrating it into a game or other software as source code or a static library, rather than using a standalone DLL. It is written in C++ and supports CMake. A hybrid kernelmode and usermode approach is available, but users must supply or make their own driver. Settings are compiled into the program, and some features need newer Windows versions. The project uses AGPL-3.0, and the README says projects using its code must also be open source.