I’ve argued before that a mechanism is most of all verified by what we can control or make happen for ourselves: e.g. we know appetite is a mechanism because we can make pills for it; chemical decomposition is a mechanism because we can do things like electrolysis; fields are mechanisms because we can make electrical generators, etc. Lo! We have a mechanical account of (some) living activities, atomic actions, and fields.
But if control is the key note or at least dominant note in defining a mechanism, we still need to narrow down the sense of control: a teacher can control the classroom without making teaching a mechanical art; and a trainer can control his horse without making it a mechanism. The control of a mechanism is not by way of incentive, persuasion, or threat, all of which take for granted that the thing you want to control has some ability to direct itself.
But self direction is exactly how the ancients and Medievals defined the living, and so the account of mechanism amounts to the idea that things are either living or machines. Plato makes a claim more or less like this in book X of the laws to prove that all of nature is being moved by the living. Plato takes this as a proof for the divine, but we are more likely to take the same premises to show that complete dominance of nature is at least in principle possible for human beings.
In this sense, the same reality we call a mechanism is what the ancients called matter – that which is, even in its being, receptive and inert. It has no self-direction but has the most perfect dominance possible by an exterior thing (and so to the extent that we consider it “objective” to leave ourselves out of accounts of things, our accounts of controlling matter must see it as having no telos or goal)