Universal Simulation Runtime

The Universal Simulation Runtime is the permanent simulation layer of GIOS: every future predictor, autonomous agent, and orchestration engine inherits this deterministic dry-run layer instead of simulating plans independently. It consumes V9's PlannerExecutionPlan (the stages, execution order, dependencies, approval requests, rollback plan, risks, and confidence model produced by the planning layer) and projects what that plan would do — a per-step predicted outcome derived from each step's confidence and reversibility, a scenario-dependent projection of which plan risks would materialize when unmitigated, a resource forecast expressed as classes and counts rather than fabricated numbers, the irreversible at-risk steps flagged as unrecoverable failure points, and an overall predicted outcome. Every step outcome cites the V9 plan step it originates from and carries the underlying cognition reference, so the dry-run is traceable all the way back through planning, cognition, and memory to the observed execution outcomes, and nothing is simulated beyond the plan. It fails closed: a missing plan yields UNKNOWN, a plan that is not itself approvable yields UNSIMULATABLE, and a plan that claims to be planned but fails the planner's own validator yields SIMULATION_INVALID — never an invented simulation. It prepares an immutable handoff for a future Prediction Runtime (V11). It never executes, never invokes a provider, never mutates memory, never changes cognition, and never changes the plan it reads — execution stays behind the Broker (→ V5) — and it inherits the A0 Intelligence Kernel lifecycle.