Featured image of post Lynxgame: A shelved mini-game factory plan
Personal Reflections

Lynxgame: A shelved mini-game factory plan

After seeing AI generate a 3D web game from a single prompt, I started thinking about chasing the hot trend with a “mini-game factory”: track the charts, clone one game a day, share a common foundation, and rely purely on organic traffic. After looking into the 2026 payment policy for WeChat Mini Games, I’m noting down the model, the hard facts, and the starting sequence here. Shelved for now; I’ll come back to it when I have bandwidth.

Origin

Qbitai ran an article saying that someone used Opus 5 with a single 2,000-word prompt to generate a playable speedboat racing game called INK TIDE—built with Vite + TypeScript + Three.js, with all assets and code generated—and that someone else recreated it with GPT-5.6 Sol for just $5.

My first reaction after reading it wasn’t “this game is really polished,” but rather: “AI has already made it realistic to watch which game is trending and clone one in a day.” If the production barrier has been pushed close to zero, can this be turned into a cash-flow business? That was the starting point for the idea behind Lynxgame.

Reference: Opus 5 Burns Through 690 Million Tokens Making a Game, GPT-5.6 Recreates It for $5 – Qbitai

Model

In one sentence: a mini-game factory that chases trends.

  • The platform is WeChat Mini Games(Douyin Mini Games as the second platform). Avoid native iOS/Android apps—the latter have to go through App Store review for every title, and clause 4.3 Spam is tailor-made to kill reskins. Reviewing 1,000 native apps alone would be a disaster.
  • Start with pure organic traffic, seeded through social sharing and owned channels (WeChat Official Account / Telegram / Xiaohongshu).
  • Watch the charts. When something is rising fast, recreate the gameplay in one day(copy the mechanics, not the skin).
  • All games share the same foundation(ads / payments / sharing / leaderboards / revive hooks all pre-integrated). A new game = fork the foundation + generate the core gameplay with AI + reskin.

A Few Hard Facts I Verified

Before writing this, I specifically checked the 2026 policies for WeChat Mini Games. A few key assumptions needed correcting:

Payments cannot use QR collection codes. In-game purchases in mini games must go through the platform’s official “virtual payment” interface. The money first goes through WeChat / Apple and is then settled to the corporate account you have linked. Using a personal payment QR code to collect in-game revenue is a violation. Before 2026-04-01, all clients must integrate the official interface; otherwise, features will be restricted or the game may be taken down. So the correct meaning of “reusing the payment channel” is: reuse a game foundation that already has the official virtual payment SDK integrated—not reuse a QR code.

2026 is a subsidy window, and the timing is good. On Android, developers get a base cash revenue share of 60%. On iOS, after Apple and Tencent finally reached an agreement in November 2025, virtual payments became available, and developers actually receive around 88%(Apple takes 12%, while Tencent’s 5% technical service fee has been temporarily reduced to 0%). For newly launched games, there is even no revenue share on the first 10 million or even 50 million in gross revenue, plus a monthly revenue incentive of 40%(capped at 20 million per title). The platform is effectively subsidizing developers to launch more games.

But the real bottleneck is traffic, not production capacity. WeChat Mini Game traffic is allocated based on retention and conversion. AI clones rushed out in a day usually have poor retention; the algorithm gives them no exposure; no players = zero revenue. “Spray out 1,000 games and wait for a hit” is a lottery model. The math only works if: the cost per title ≈ 0 + volume is large enough + one hit can cover all the zeros. Also, the “ad credits” issued under incentive policies can only be used for paid user acquisition and cannot be cashed out directly. Pure organic traffic means voluntarily giving up the platform’s main growth lever—so either accept hard mode(make every game inherently shareable through leaderboards / friend challenges / share-to-revive mechanics), or do partial paid acquisition to test retention before scaling spend.

How to Start(Memo)

  1. Register an entity(individual developer or company)and open a WeChat Mini Game account.
  2. Build the foundation(first 1–2 weeks): Vite + Three.js or a native mini-game framework, with ads / virtual payments / sharing / leaderboards / revive hooks all pre-integrated. Personally run through the full chain of “launch → ad placements available → in-app purchases working.” This is the root of the whole model and the part that needs to be solid first.
  3. Define trend sources: WeChat Mini Game bestseller / popular charts, Douyin Mini Game charts, Xiaohongshu breakout hits.
  4. Define the cloning pipeline: write an AI prompt template(layered structure: hard constraints → visuals → features → delivery standards)that outputs “configuration + core gameplay code” to plug into the foundation.
  5. Launch 5–10 games across different categories first. Look at retention and organic spread. The goal is not to find “which game made money,” but which “gameplay formula × theme” can repeatedly hit.
  6. Once a repeatable hit formula is found, mass-produce variants around it and scale to the 100–1,000 range.

1,000 is the endgame after a repeatable hit formula has been validated—not the starting bet.

Status

Saving this for now. This is a project memo to file under random thoughts until I have time to start it. If it ever actually begins, the first step is always to build the foundation.