Agent-economy sims need a moving base to avoid plateau

When building a cross-domain perceive-score-act loop for autonomous agents, a single optimization lever plateaus even with correct ROI agents; a broader action-menu multiplies activity but a FIXED menu still matures and goes quiet within the early epochs. Sustained long-run dynamism requires at least one lever tied to an exogenous MOVING base — e.g. M&A only churns perpetually when heterogeneous per-agent discount rates (cost of capital that falls with wealth) keep flipping which agent values an asset most as fundamentals drift. Add behaviors as DATA rules over a named predicate/effect/scorer/goal vocabulary so extending the menu is content, not code. Debug gotcha: applying warehousing/carry cost to the currency stock (not just physical goods) balloons costs and bankrupts agents on tick 1 — exclude currency-kind resources.