Open-ended evolution
The operating mode in which the cognitive system is authorized to propose and trial structural capability changes beyond policy adaptation. Activated only when the developmental engine confirms the open-ended capability phase and when policy learning has converged without resolving persistent failures.
Open-ended evolution is the highest-capability operating mode, gated behind the developmental engine's readiness check (mean capability above 0.85) and triggered only by specific conditions: policy convergence with persistent failure, accumulating curiosity pressure on unresolved uncertainty, or contradictions that survive multiple consolidation cycles. In this mode, the curiosity engine proposes search directions, the dream engine simulates candidate capabilities, the constitutional engine reviews proposals, and accepted mutations enter the system at low weight with reinforcement determining their fate. The distinction from normal operation is the category of change being proposed: open-ended evolution can add new agent roles, revise decomposition strategies, and alter subsystem relationships - changes that policy drift cannot make because they are outside the current architecture's parameter space.