strict old-skill vs new-skill comparison

ui-polish ยท same task A, real old vs new outputs

This comparison uses one concrete task A and two real outputs produced at different times in the update wave: the old side was created before the skill update, and the new side was created after the skill update.

Method: Task A is the same floating appearance control panel, and the chip cluster still uses uneven label widths (Quiet, Studio Mode, Deep Focus). The old side already has platter travel, local relief, morph continuity, lens clarity, and settle grace, but the resting lens still feels slightly broad and anonymous on Quiet. This update adds rest-lens-anchoring.md, which keeps the same grouped field but makes the quiet resting lens read more like a clearly held home state on the default chip.
First compare the resting Quiet state. Then hover Studio Mode and move out of the cluster. Compare whether the field returns to a vague quiet haze, or to a more clearly anchored home on Quiet.

Old skill on task A

pre-rest anchor
Old home state: the field already returns gracefully, but the resting lens still feels a little broad and anonymous. The cluster settles well, yet the default chip is not held quite clearly enough as the quiet home point.

New skill on task A

+ rest-lens anchoring
New home state: the same field still travels, relieves, and settles gracefully, but now the resting lens tightens into a clearer home anchor on Quiet. Hover feels more like leaving and returning to one known object instead of fading into a generic premium haze.
Task A: Polish a tiny floating appearance control panel for a persistent Ubuntu desktop utility without changing features, using an uneven-width scene cluster so the resting state must prove whether the field has a clear home anchor as well as graceful hover behavior.
Why this update matters: once settle grace already works, the next Apple-like refinement is often not more motion but a quieter, clearer resting home for the same grouped material.