mantle

Climbing onto a ledge that is too high to step onto. The pure decision and the pure path: given what a probe saw in front of the character, is there a lip worth climbing, and where does the body travel to get on top of it? Babylon-free, deterministic, unit tested.

It is not a water feature

It arrived as one — Tonio, swimming: "when you swim to the water's edge, you just pop instantly to the surface onto the land." But climbing out of a pond is mantling a ledge of height h, and the water is incidental. Written as swim_exit it would be one animation that only ever plays at a shoreline; written as a mantle it is the bank, the low wall, the crate and the ledge, from one verb. Quaternius agrees, and says so in its naming: the clips are ClimbUp_1m and ClimbUp_2m — indexed by HEIGHT, not by what you were doing before.

That is also the MOBILITY-DESIGN rule doing its job. A ledge is a fact about geometry; _climbable would be a fact about a level designer, and the design doc calls that out by name as the thing that makes players stop believing their eyes.

The band, and why both ends of it matter

A mantle is what happens BETWEEN two things that already work:

So the lower bound is not a tuning choice, it is whatever the walking code already absorbs — pass it in rather than restating it, or the two disagree and you get a band of heights that is neither steppable nor climbable and reads as an invisible wall.

The path is an ARC, and a lerp is wrong

Straight-lining from the water to the ledge top drives the body THROUGH the lip. A mantle goes up first, then in — that is what the shape of the animation is — so the path rises early and translates late, with an overlap in the middle so it is one motion rather than two.

Timing is the one thing here that is a choice rather than a measurement, and it is expressed as fractions of the clip so it retimes itself when the animation set changes.