The current MaOS prototype is free and already keeps the final decision with the player. The AI maintains the operational surface — rules, state, risks, resources, and consequences — but it should not quietly turn into an autopilot.
The next design question I am documenting is whether an advisor can go further without taking control. A future mirror-style advisor could transparently reflect patterns in the player's own decisions: what they protect first, which risks they accept, and what they repeatedly postpone. The purpose would be to help the player see their strategy, not to prescribe a universal best move.
This adaptive idea is not a shipped feature yet. It needs careful testing, an always-available neutral mode, and clear disclosure of what the system inferred and how uncertain that inference is.
One-time support helps fund structured testing, documentation, proof-of-play screenshots, editorial review, and future levels. The public prototype remains free:
https://temnik.itch.io/maos