THE WALL CHESS PETRI DISH

We built a topological strategy engine in raw HTML5. Now, we need your human intuition to evolve it. No cloud. No tracking. Just pure, adversarial growth.

In the subscription economy, you rent your mind. At Mnemosynes, we build Sovereign Assets. Every app we ship—including this simulator—is a zero-trust, offline-first entity.
​We used AI as a high-bandwidth compiler to build this state machine. But while the code is fast, the 'brain' is currently that of a bacterium. It understands pathfinding (BFS), but it doesn't understand deception.
​Play against it. Break its logic. Help us map the topology of strategy.

Step 1: The Duel. Play against the V1.0 engine. Use the "Simulation Mode" to watch the AI battle itself.
​Step 2: The Stress Test. Play with your children. Their non-linear moves are the ultimate "Evolutionary Pressure" for a path-finding algorithm.
​Step 3: The Data Drop. Use the [COPY QGN] button to capture your session logs.
​Step 4: The Contribution. Paste your QGN logs in the comments or send them to us. Every log is used to train the neural network for Version 2.0.

We don't just build games; we build Data Pipelines.
Architecture: 100% Client-Side Vanilla JS/HTML5. Zero server dependency.
Logic: Breadth-First Search (BFS) pathfinding verified for topological legality (Diagonal Jumps included).
Dataset Ready: Generative engine capable of producing the 3 \times 17 \times 17 tensors required for deep Convolutional Neural Network (CNN) training.
Sovereign Mode: Once unlocked, you can save the page as a single file and play it offline—forever.

Built on the Mnemosynes Standard.

If we can map the 289-node topology of a strategy game in an afternoon, imagine what the Mnemosynes OS can do for your corporate network and legacy data.

Download this version to be able to save game data.