Skip to main content
Anti-cheats fingerprint menus by the Roblox asset ids they reference, and by probing whether a known id is loaded with ContentProvider:GetAssetFetchStatus. Secure mode makes sure nothing loadable can be traced back to Rayfield. Turn it on by setting a global before you load Rayfield.

​
Why there is nothing to swap

Gen2 builds its whole interface in code rather than loading a Roblox model, so there is no asset id to swap out and nothing like RAYFIELD_ASSET_ID to set. The only detectable surface is icon images, and secure mode handles those two ways.
  • Built-in icons (chrome, dropdown ticks, and the rest) are downloaded once and served from a local file through getcustomasset, so the UI still looks normal.
  • Your icons, meaning any raw asset id including Lucide-style ids, are blocked and render blank. Referencing them by id is exactly what gives a menu away.

​
Keeping your own icons

Cache them yourself and pass the getcustomasset result instead of an id.
The net effect: in secure mode the only assets Rayfield ever loads are local getcustomasset files, whose content ids are per-machine and cannot be guessed or probed for. No marketplace id and no thumbnail is ever loaded, so GetAssetFetchStatus on anything tied to Rayfield comes back empty.

​
Avatars still work

Instead of an rbxthumb:// id, a headshot is fetched once through Roblox’s thumbnail API and cached to a local file, exactly like icons. Use avatar = userId for this. A raw rbxthumb:// string you pass yourself is blocked.

​
It also goes quiet

Secure mode makes no warn or print calls, since console output is another thing anti-cheats watch. Config and callback errors that would normally log are swallowed.
Develop with secure mode off, then switch it on for release. Because errors are swallowed, a mistake is much harder to spot while it is on.
Secure mode needs an executor with file access (getcustomasset, writefile). If some built-in icons cannot be cached, the window shows a one-off notification so you know a few will be blank.