Use a game prototype maker to test a mechanic before you pick an engine
Most game ideas sound fun and play flat. The cheap way to find out is a rough, playable build of the core mechanic before you commit months to an engine and art. Anticode can work as a game prototype maker for that first step: describe the mechanic, play it in the browser minutes later, change one variable at a time, and keep recent versions so you can go back to the one that felt right.
Prototype the verb, not the game
A prototype answers one question. "Is it fun to swing between moving anchor points?" can be answered in a day. "Is my open-world grappling adventure fun?" can't. Write your question down, and build only what it needs:
- The core action, with its feel: speed, acceleration, how input maps to movement.
- One small space to do it in, designed to show off the action.
- A goal and a failure, so play has stakes: reach the flag, don't fall.
- A restart button, because you'll press it hundreds of times.
- Nothing else. No menus, story, progression or art beyond what makes the action readable.
Expose the numbers
Game feel lives in numbers: jump height, gravity, coyote time, friction, enemy speed. Ask for a small tuning panel on screen with sliders for the five or six values that matter, so you can adjust while playing instead of asking for a rebuild each time. Note the values when something feels good, then ask for them as the new defaults.
| Value | What it changes |
|---|---|
| Gravity and jump velocity | How floaty or snappy jumps feel |
| Coyote time | How long after leaving a ledge a jump still works; forgiving controls feel better |
| Jump buffer | Whether a jump pressed just before landing still counts |
| Air control | How much the player can steer mid-air |
| Camera lag | How tightly the view follows the player |
Describing the prototype
Be precise about input and feel: "A game mechanic prototype: a 2D side-on grappling hook. Click on an anchor point to attach, swing with momentum, release to fly. Anchors drift slowly. One screen-wide test level with a gap of increasing width and a flag at the end. Arrow keys and mouse. A tuning panel with sliders for rope length, swing gravity, release boost and anchor speed. Flat shapes only, a timer and a restart key."
Anticode writes it as one app, drawn in code on a canvas, with a start screen so the keyboard reaches the game after a click. When Cloudflare Browser Rendering is available, new app builds are also opened in a headless browser, screenshotted and clicked through, and errors found that way are fixed before you see it.
Versions as an experiment log
Version history keeps your saves as versions, up to a number set by your plan, and you can preview and restore any of them. Use it deliberately:
- Change one thing per request: "anchors stop drifting", "release boost doubles".
- Play each version for a few minutes and jot down how it felt.
- When a change makes it worse, restore the previous version rather than asking for the change to be undone in words.
- Keep a short note of which version number had the best feel, and why.
Playtesting
Publish the prototype and send the link: anyone with it can play in their browser, on a laptop or (if you've asked for touch controls) a phone. Watch people play if you can; silence and restarts tell you more than written feedback. For remote testers, ask for a short feedback form at the end that sends their name, email and comments to you through Anticode's form pipeline. Scores and progress stay on each player's device, so there's no automatic gameplay data.
When to move on
A browser prototype isn't the final game, and some things can't be tested here:
- No online multiplayer. The sandbox has no network. Local two-player on one keyboard works; netcode doesn't.
- No downloaded assets. Art, sound and models from other sites can't load; everything is drawn and synthesised in code, or uses your own uploaded images.
- No console or controller certification work, and gamepad support needs testing in your browser.
- No engine project. You can export the prototype as one HTML file, but nothing converts it to Unity, Godot or Unreal. Rebuild there once the mechanic is proven, using your tuning values.
If the prototype becomes a small browser game in its own right, see browser game maker and, for a specific genre, platformer game maker.
Questions people ask
How rough should a game prototype be?
Rough enough that you're not attached to it. Coloured boxes are fine; the question is whether the action is fun.
Can I prototype a 3D game?
Simple 3D can be drawn with WebGL in the browser. Keep the scene small, and test performance on the devices your testers use.
Can I protect the idea while playtesting?
Published apps are public to anyone with the link, with no password option. Share the link only with people you trust, and unpublish when testing ends.