Sounds like an interesting project, but I think it's important to find a good way to interact with browser APIs. They all work differently, like rAF, geolocation, pointer lock, camera access and so on. I think relying on the unsafe mode for all that could lead to a frustrating experience but also implementing those APIs in a different way would be weird and you have to keep up with new APIs and changes browsers introduce. I suggest having an idea on how to solve this before spending a lot of time developing the actual language. If it limits what I can do with JS I would not use this and just stick with Typescript.