Analysis updated 2026-05-18
Audit a published game or your own design document for balance problems using graph theory computation.
Generate theory tested redesign proposals for a game, verified against the same math engine.
Run a blind re-audit in a fresh session to independently check a proposed redesign.
| apetrovcode/game-design-skills | 0xustaz/streamgate | a-bissell/unleash-lite | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | Python | Python | Python |
| Setup difficulty | easy | hard | hard |
| Complexity | 2/5 | 4/5 | 4/5 |
| Audience | developer | developer | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires Claude Code and Python 3 with no extra packages to install.
game-design-skills is a pair of add ons for Claude Code, the command line coding assistant, that turn a chat conversation into a tool for analyzing and improving game designs. The first skill, game-analysis, examines any game, whether it is a published title, your own design document, rules text, or prototype code, and reports what kind of game it is, what makes it enjoyable, and what parts of its design might be broken. The second skill, game-redesign, takes that audit and proposes bold new versions of the game, including new mechanics and rebalanced interaction charts, then checks each proposal against the same underlying math before handing back a specification meant to be reviewed again from scratch. What sets this apart from a simple prompt is that both skills route their important balance claims through a small, dependency free Python script that actually computes things rather than guessing: it detects cycles in how game choices beat each other, finds strategies that are always worse than another option, searches for combinations of choices that together cannot be beaten, and calculates viability scores using a technique called eigenvector centrality. So when the tool claims there are no dominated strategies, or that a certain balance property holds, that is a result you can rerun and verify yourself using the included examples, one of which walks through classic rock paper scissors as a demonstration. The project includes a folder of real, unedited example runs showing the full loop of auditing a game, proposing redesigns, and then having a separate, memory free session independently re-check the proposal, including cases where the re-audit catches a real design gap the original session missed, and a case showing the tool correctly declines to force a mathematical structure onto a game like Tetris that does not have one. Installing it requires Claude Code and Python 3 already on your machine, since it uses only Python's standard library. The knowledge base behind the skills is meant to be swappable, so someone could replace the built in design theory notes with their own framework. The author is upfront that the tool can prove a design is not mathematically broken but cannot judge whether a game actually feels fun, so playtesting still matters. This project is aimed at game designers and developers using Claude Code. It is released under the MIT license.
game-design-skills adds two Claude Code skills that audit any game's balance using real graph theory computation, then propose and verify redesigns.
Mainly Python. The stack also includes Python, Claude Code.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.