← Back to the blog
15 July 2026

I'm rebuilding a dead game for an audience of two

I'm rebuilding a dead game for an audience of two

Back in 2017, my husband and I got completely hooked on a mobile game called 7 Paladins. In 2018, they shut the servers down. We've been looking for something that felt the same ever since, and nothing's come close.

So I'm going to build it back myself.

Quick bit of background. Before I did web stuff, I made games. That was my actual job for a few years. Now I build for the web and write here about doing it with AI. This project is where those two things finally overlap. And it only feels doable because of the AI tools I already use every day.

What was 7 Paladins?

It was a 3D hero-collection RPG with MOBA-style combat. You collected characters, leveled them up, built teams, and fought your way through a campaign, boss fights, guild battles, and a PvP ladder. Fifty-seven heroes, each with their own skills. A small studio published it around Southeast Asia, and (this part I only found out while digging) it actually started out as a Chinese game.

It was never a big famous title. Didn't need to be. It was ours.

It was also the kind of game that only works while someone keeps the servers running. No servers, no game. And that's exactly how we lost it. The company moved on, switched things off, and a game we'd sunk a year of evenings into was just gone. Just gone.

Why rebuild something nobody else misses?

Because that's kind of the whole point. Nobody's asking for this game back. I just want it back, for the two of us.

Let me be clear about what this is and what it isn't. I'm not selling anyone's game. I'm not putting it back on an app store. It's a private thing for two people who miss it, running on our own little server. That's the entire plan, and keeping it that small is what makes it actually possible (and legal haha).

I'm also being realistic about it. I'm not rebuilding some huge live game for thousands of players. Single-player first. Then a private server just for us. And the multiplayer stuff, the ladder and all that, I'll fill in with bots. Fun fact: DC Legends, another one we lost, faked a bunch of its "players." So the bot thing is kinda accurate.

Where the AI actually fits in

Not the way people assume. I didn't type "make me a game" and go get a coffee.

What AI is genuinely great at here is the tedious stuff that would normally make a solo project like this a non-starter. Reading ancient file formats (some in Chinese). Picking apart a years-old app. Writing little scripts to pull things out. Keeping track of a plan spread across hundreds of files. I still make all the actual decisions. The AI just makes the boring parts fast enough that I don't give up halfway.

That's the honest version of "AI changes everything." It doesn't do the work for you. It just gets you past the point where you'd usually quit.

What I've figured out so far

This is the part I didn't expect to enjoy this much. All I started with was one 200MB app installation file I recovered. Everything else was hidden inside it, and we've been pulling it apart piece by piece.

Here's what's turned up so far:

  • The game runs on an old version of Unity. That's good news, because it means the game's code was written in something readable. Mostly.
  • "Mostly," because the actual code is encrypted. Locked up behind an anti-piracy wrapper. You can see the file, you just can't read what's inside it. That's been the big plot twist of this stage.
  • The art wasn't locked, though. So I grabbed all of it. Every one of the 57 hero models, the textures, the audio, the UI, and 1,270 animation clips. Every attack, idle pose, and death animation, all pulled out clean.
  • Here's a nice surprise. Even though the code itself is locked, the list of everything in it wasn't. So I now have the names of around 800 pieces of the game. The combat system, the skills, the guild raids, the PvP ladder. I know every system that existed, I just don't know exactly how each one works yet.
  • And an origin story I wasn't expecting. All the internal names are in Chinese. Sun Wukong, the Monkey King, is in there. So is the game's original title, 圣光契约, which I never would've found otherwise.

What's next

One piece is still locked away. That encrypted code holds all the numbers that make the game feel like the game. The stats, the skill formulas, the balance. There's a way to get at it (you catch it while it's briefly unlocked in memory as the game runs), and that's the next thing I'm going to try.

After that, it's rebuilding the whole thing in a modern engine, one phase at a time. One working battle first. Then a full single-player game. Then the server. Then the bots. I'll write about each step here as I go, the wins and the parts where I get stuck, because there'll definitely be parts where I get stuck.

Why I'm even telling you this

Not because it's useful. It really isn't, to anyone but two people.

I'm telling you because it's the clearest example I've hit of something I keep going on about with these tools. The bar for doing ambitious personal projects just dropped way down. That thing you gave up on years ago because you didn't have the time or a team for it? A lot of those are quietly doable now.

Mine was a game. We're getting it back, one phase at a time.

Want a site like this?

Take a look at the packages, or just say hello.

See the packages →