We propose a unified conformal prediction framework for infinite-horizon policy evaluation that seamlessly accommodates both on-policy and off-policy scenarios.