#Decision Theory MCP Servers

Charlie Munger spent 60 years proving that a man with one model is a man with a hammer. Every problem looks like a nail. An AI proposed a strategy without inverting it, without tracing second-order effects, without mapping incentive conflicts, without admitting what it does not know, and without a margin of safety for being wrong. That is not analysis. That is a first draft dressed as a final answer. This tool forces five Munger-level mental models: inversion, second-order effects, incentive architecture, circle of competence, and margin of safety.

Every strategy proposed by an AI treats the world as a single-player game. No opponents. No counter-moves. No equilibrium analysis. The AI said 'our competitive advantage' without mapping a single opponent's payoff, analyzing their best response, or checking if the strategy survives rational counter-play. Nash proved in 1950 that every finite game has at least one equilibrium. Ignoring this is not a strategic choice, it is a mathematical error. This tool forces five game-theoretic axes: payoff mapping, equilibrium analysis, information structure, mechanism design, and repeated dynamics.