To see that causally closed systems need not become forever open once an agent interferes with the system externally, suppose I run a program on an isolated computer that performs a simple, but unending loop. That instantiated, looping program is a causally closed system.
Suppose I stop the program by turning off the computer. When I shut the computer off, I acted as an agent from outside the causally closed system. From the system’s perspective I might as well have performed a “divine miracle”.
Suppose I start up the computer again and also start the looping program. The looping program has not been affected by my "divine miracle" of stopping the computer earlier. The looping program is now just as closed as it originally was.
Plantinga makes an important point about the idea of causally closed systems and physics in the book you cite, Where the Conflict Really Lies, page 79:
...it is no part of Newtonian mechanics or classical science generally to declare that the material universe is a closed system. You won’t find that claim in physics textbooks--naturally enough, because that claim isn’t physics, but a theological or metaphysical add-on.
Being causally closed simply means being “subject to no outside causal influence”. We can assume systems are causally closed to simplify an analysis of them. We don’t know that any systems are actually causally closed unless we have set them up that way, such as computer programs or experiments.