🍄 Nintendo2001HAL LaboratoryVERY HARD

Super Smash Bros. Melee rebuild deck

VERY HARD

You will get a good prototype. The product is what comes after.

This moved from NO to KINDA on review and it should have. Two fighters on one stage is a small honest build: the knockback formula is public, the simulation is a fixed 60Hz loop, and the frame data is a JSON table. HAL shipped the original in thirteen months. Expect the first draft to test hitboxes at frame boundaries, and expect that to look completely fine until you halve a move's startup, at which point attacks start passing through people. The other review priority is the knockback function, where an agent will invent something plausible that behaves at 40% and is nonsense at 150%. What you will not reach is the game people actually play, which is twenty-six characters, rollback netcode, and a set of physics bugs nobody designed and nobody has ever successfully prompted for.

Difficulty: VERY HARD · 4 of 5
First playable: A project
The original: 60 team-years
Released: 2001 · Nintendo
Genre: Platform Fighter

The one-shot build prompt

Build a local two-player platform fighter. Canvas 2D, fixed 60Hz simulation, two characters, one stage. No roster, no netcode, no single-player.

Requirements:
- Percentage damage, not health. Knockback = (a damage-derived term times the move's knockback growth, plus its base knockback) divided by defender weight, and hitstun frames derive from the resulting knockback. Everything downstream depends on this: one function, with unit tests.
- The load-bearing detail: hitboxes and hurtboxes are capsules swept from last frame's position to this frame's, tested capsule-against-capsule, never as point-in-box at frame boundaries. A fast attack at 60fps travels further in one step than its own hitbox is wide and will tunnel clean through a defender otherwise — and it will look correct until the first time you speed a move up. Give every hitbox an ID so a multi-hit move cannot rehit the same target.
- Frame data as data: every move is { startup, active, recovery, damage, base, growth, angle, hitboxes[] } in a JSON table, with a debug overlay drawing every box and the current frame index.
- Hitlag proportional to damage, freezing both fighters, then launch.
- Shield with decay and shield stun, grabs that beat shields, and an air dodge that carries directional momentum.
- Off-stage play: no floor, a blast zone, one double jump, and one recovery move with a fixed distance. Dying means leaving the boundary.

Definition of done: I hit a 120% opponent with a fully charged smash, they leave the screen, and the debug overlay shows the swept capsule that connected.

Before you start🧰 The toolkit — models, MCP servers, art and 3D

What you lose

Twenty-six characters with roughly thirty distinct moves each, on hand-tuned frame data people are still arguing about twenty-five years later
Rollback netcode, which fans shipped in 2020 — nineteen years after release — and which is harder than the game
Readability at 60fps: silhouettes, hit effects and freeze frames that tell you what just happened in three frames
The accidents — wavedashing, L-cancelling — emergent from physics quirks nobody designed and no prompt will ask for

What to steal

Directional influence — the defender's stick input during hitlag rotates their own launch trajectory.

It makes getting hit an input rather than a pause. The attacker still wins the exchange; the defender still has a decision inside it. Any game where something knocks you back — a brawler, a shooter with explosions, a racer with contact — gets deeper the moment the receiving end is handed a steering wheel.

ScopeAn evening— build just this, not the game

Build a single-file HTML page, Canvas 2D, one attacker and one training dummy, demonstrating directional influence.

Requirements:
- Press a key to launch the dummy: a base launch angle, a knockback magnitude, then both entities freeze for hitlag frames proportional to damage.
- During hitlag only, read the defender's arrow-key input. Project it onto the launch vector: the perpendicular component rotates the trajectory by up to 18 degrees, the parallel component does nothing at all. That is the whole idea — the defender cannot escape, but they choose where they are going.
- Run the same hit twice side by side, with and without full perpendicular input, drawing both arcs as persistent trails.
- Put a blast boundary on the right, positioned so correct influence survives a hit that otherwise kills.
- Show the input window as an on-screen countdown. A mechanic the player cannot perceive is not a mechanic.

Definition of done: I survive at 140% by holding the right direction during a five-frame window, and die at the same percent when I do not.

Make it different

Not a clone — Super Smash Bros. Melee re-themed and cut down to something you can finish. The angle is the important reel: it is what turns a project into a weekend. 24,360 ways to land.

Velvet Rope One Run

Super Smash Bros. Melee, but a roguelike run instead of a campaign, set in a heist-movie glam night. Rendered in CRT arcade with scanlines.

Themea heist-movie glam night

Anglea roguelike run instead of a campaign

LookCRT arcade with scanlines

Build a browser game called "Velvet Rope One Run".

The pitch in one line: a platform fighter in the spirit of Super Smash Bros. Melee, but a roguelike run instead of a campaign, set in a heist-movie glam night.

WHAT THIS IS
Take the SHAPE of Super Smash Bros. Melee — its core loop, and whatever makes that loop good — and rebuild it as your own game in a different world at a smaller scope. Use your own names, art and audio, which you want anyway: a reskin of somebody else's game is less interesting than a new one that works the same way.

THE ANGLE — this is the important constraint, honour it above everything else
A roguelike run instead of a campaign. Twenty to forty minutes, procedurally assembled from hand-authored chunks, permadeath, and a meta-progression drip. Replaces authored content volume with combinatorics.
This is what makes the project finishable. If a decision would push the scope back towards the original's, take the smaller option every time.

THE THEME
A heist-movie glam night. Casinos, tuxedos and a vault. Style is mandatory; competence is what keeps the style from getting you caught.
The theme is not a coat of paint. Let it change what the mechanics mean — a reload, a health pack and a locked door should all be things that make sense in this world and nowhere else.

LOOK
CRT arcade with scanlines. Post-process pass with scanlines, barrel distortion, phosphor bloom and slight chromatic aberration at the edges. Keep it subtle enough to read text through.

REQUIREMENTS
- Runs in a browser with no build step. Canvas 2D unless the angle genuinely requires 3D, in which case three.js.
- Fixed-timestep update loop with interpolated rendering, so it behaves the same at 60Hz and 144Hz.
- Every tuning constant in one CONFIG object at the top, exposed as live sliders in a debug panel. You will find the feel by dragging those, not by prompting.
- All content — levels, entities, balance numbers — in JSON or plain data files, never inline in the logic.
- Audio generated with the Web Audio API rather than asset files.
- Respect prefers-reduced-motion: keep the fades, drop the shake and the parallax.

BUILD ORDER
1. The core loop with placeholder rectangles, at the reduced scope the angle demands. No theme, no art, no audio. Make it fun as rectangles first.
2. The angle's consequences. Cutting to bots, or to turn-based, or to one level changes the design rather than just shrinking it — find out how and rewrite step 1.
3. The theme, as content and data.
4. The look, applied last as a rendering layer over a game that already works.

Start by writing a one-page plan: the core loop, what the angle removes and what that frees you to do properly, and what you are deliberately leaving out. Wait for me to approve it before writing code.

Prior art — read before you build

Half the time one of these already does what you wanted, and the other half it saves you a week of solving a problem someone documented in 1997.

Did you build it?

0Share on X

One count per game. No account — it is tied to a salted hash of your IP, which is not stored in a form anyone can reverse. Every build adds 60 team-years to the counter on the homepage.

Then try these

Or roll something nobody has built yet →

Questions

How hard is Super Smash Bros. Melee to rebuild with an AI agent?

Yes — every game is buildable at some scope. The question is how hard, and Super Smash Bros. Melee is VERY HARD (4 of 5). You will get a good prototype. The product is what comes after. The knockback formula is public and two fighters on one stage genuinely lands, but the game people actually play is twenty-six characters of hand-tuned frame data plus rollback netcode fans needed nineteen years to ship. Reckon on a project to something you can actually play — not to something that matches the 60 team-years HAL Laboratory spent.

What makes Super Smash Bros. Melee hard to rebuild?

The knockback formula is public and two fighters on one stage genuinely lands, but the game people actually play is twenty-six characters of hand-tuned frame data plus rollback netcode fans needed nineteen years to ship. That is why it sits at VERY HARD rather than a tier either side of it. If you want the idea without the project, the transferable part is directional influence — the defender's stick input during hitlag rotates their own launch trajectory. — an evening of work.

How long does it take to build a Super Smash Bros. Melee clone?

Wrong question, and it is worth saying why. Nobody rebuilds Super Smash Bros. Melee — the original took HAL Laboratory on the order of 60 team-years, released in 2001 on Nintendo. What you can do is reach something playable and genuinely yours in about a project, and then keep going for as long as it stays interesting. Any figure that claims otherwise is measuring a prototype and calling it a game.

What do I lose by building my own Super Smash Bros. Melee?

Twenty-six characters with roughly thirty distinct moves each, on hand-tuned frame data people are still arguing about twenty-five years later. Rollback netcode, which fans shipped in 2020 — nineteen years after release — and which is harder than the game. Readability at 60fps: silhouettes, hit effects and freeze frames that tell you what just happened in three frames. Those are the parts a prompt will not hand you.

Which AI coding agent should I use to build it?

Any of Claude Code, Codex or Cursor will handle this prompt. Paste it as your first message, let the agent scaffold the project, then iterate on feel — the second, third and tenth prompts are where a game actually gets good. Work in an empty folder or a fresh git branch so you can throw it away cheaply.

Get the next game

One email when new games go on the list. No other reason to email you, ever.