Not sure if there's still any interest here, but here's my 2¢
This "Construct2" doesn't sound like it makes sense for something like this. It looks like GameMaker: already a streamlined rapid development tool for prototypes or simple games. You might try to put together a sorta decent-ish standalone Cave Story clone, but making another, more specialized editor within this type of editor sounds implausible, like just way too many layers of abstraction.
If something like this were to be actually attempted, it would probably make sense to use
NxEngine-Evo as a starting point. You'd already have reasonably accurate-to-the-original code for the basic physics, control of the player character, weapons, enemy AI, graphics, sound, etc.
And it can even extract all the graphics resources and ORGs from an existing cave story installation, allowing it to re-use Cave Story's sprites, tilesets, ORG music, etc. without actually bundling them (thereby avoiding the questionable legalities that would come with doing so).
It would still be a lot of work even re-using all of this, but you wouldn't have to totally reinvent the wheel.
If we remove the online co-op as a design goal, it "simply" (using that term loosely) becomes a matter of developing an editor that presents a more "abstract"/"high-level"/"user-friendly" interface than existing tools like Booster's Laboratory.
This editor could either output to TSC and the existing map data format behind the scenes, or it could develop its own.
- Developing its own would require Nxengine to be used and would require significant changes to it to support the new map format.
- Re-using the old format. This could possibly even be used to make mods that run with the original Cave Story engine, but even then, to get the type of rapid testing that's usually desired from this type of application it would better if the engine were integrated into the editor, which is probably only plausible when using the open-source clone.
I do suspect that a new (or at least altered) underlying map data and script format would be required to get exactly the type of behavior desired for something like this; trying to make truly "Mario Maker"-style behavior work within limitations of vanilla TSC and the vanilla engine might greatly limit what is possible, considering the existing map and script format is designed for low-level editing.
And if this or other significant changes to the base engine are required, then NxEngine-Evo as a base is the most realistic choice. Not to downplay all the progress that's been made over the years in terms of hacking doukutsu.exe directly, but editing actual source code (even if it's C++) will always be easier/more flexible than writing Assembly and injecting it into an existing program.
Once again the limiting factor is that this would still require someone with reasonably good programming skills to put in a fair bit of work. But if you really wanted to do it, I think this approach would make the most sense.
EDIT: I'm kinda sorry for bumping this post (even slightly) considering that it's probably off-topic to begin with in a subforum that's meant to showcase completed (or at least functioning-WIP) mods, but I did feel I had something to say about this.
EDIT 2: Wowie wow, looks like someone is already adding online multiplayer to (a fork of) NxEngine-Evo, called "
NetXEngine". Using that as a base, I guess it's plausible this hypothetical new editor
could support online multiplayer? I honestly didn't see that coming, even the more "wishful thinking" sounding features of this idea are starting to look plausible. Open Source for the win, I guess.
Of course, there's still the problem of, y'know, actually
making this "Mario Maker style" editor, regardless of whatever other features might be reasonable to add to it.