An AI tower defense maker where the numbers are yours to set

In tower defense, almost every design decision is spatial. How long the lane is. Where it narrows. Whether there is a second path that opens on wave nine. How much buildable ground sits beside the choke point, and whether a player who commits everything to it gets punished later. None of that is a sentence you type. It is a shape you draw.

So that is the part you do yourself. In Stage Engine you lay out the terrain, the lane, the buildable plots and the props by hand, in a builder that works like a level editor. You can walk the lane at the scale a creep walks it, which is the only reliable way to find out that your beautiful S-bend is three times longer than it looked from above.

This is why a prompt-only tool struggles with the genre. Describing a map in words and hoping the result is balanced is slower than drawing it, and the feedback loop is worse. You want to move a rock four metres and look again.

The code for a tower defense game is not conceptually hard, but there is a lot of it, and it is the sort of code that breaks in boring ways. Pathfinding that reroutes when a player walls off a lane. Wave spawners with per-wave composition. Towers with range, fire rate, damage type, targeting priority and an upgrade tree. Currency, sell values, interest on unspent gold if you want it.

Director 1 writes that as GDScript inside your project. You describe what you want in ordinary language: slow tower that stacks up to forty percent, ignores armoured enemies, and costs more each time you place one. It writes it and you play it.

You will not be asked to read the script. You will be asked whether wave twelve feels like a wall or a difficulty spike, which is a judgement only someone playing it can make.

Every tower defense game lives or dies on its numbers, and no tool can guess them for you. What a tool can do is make each pass cheap. Change the creep health curve, press play, watch wave seven, change it again. Ten minutes per iteration kills a design. Ten seconds does not.

Because the game runs locally on the copy of Godot bundled with the app, pressing play is immediate and the framerate is your machine's, not a server's. That matters once you have sixty enemies pathing at once, which is a modest wave by the standards of the genre and a heavy one by the standards of a browser tab.

When something is off, you say so in plain terms. Splash damage feels too generous. The early game is boring for the first three waves. Director 1 revises the code and you go again.

The reason to build tower defense in a real 3D engine rather than a genre-specific template is that you are not restricted to the template's idea of the genre. Towers that reposition. Enemies that dig under the lane. Weather that changes tower range, because the world underneath supports weather. A lane that floods on a timer and forces a different build.

None of that is a feature you unlock. It is a consequence of the project being a normal Godot project with code written for it on demand rather than a fixed set of behaviours you pick from a menu.

It also means the game is yours in a practical sense. The files sit on your disk, in the format Godot reads, and nothing about them depends on Stage Engine continuing to exist.

It is not a browser tool. Stage Engine is a desktop application for Mac and Windows that you download and install, and there is no in-tab editor or preview. If what you wanted was a tab you could open on a school computer, this is not that, and pretending otherwise would waste your afternoon.

It is also not offline. Director 1, asset generation and world data come over the network, so you need a connection while you build. Playing the game you built does not need one.

And it is not a one-click genre generator. You will make a map, tune the numbers, and decide when it is fun. That is the work. What is removed is the scripting, not the design.