No. Use any retargeting tool you like, or Blender's own Copy Rotation and Bake. RetargetProof checks the result against the source and tells you how far it is, on which bone, on which frame, and why.

Any: it reads the evaluated pose of the two armatures, not a tool's settings. It was run on results made with Copy Rotation and Bake, RetargetKit, Rokoko Studio Live, KeeMap and Expy Kit.

No. Degrees are compared bone by bone, millimetres are measured on the target's own skeleton, and feet, floor and hips are compared with what the source itself does, scaled by the two rigs' sizes. The distance between the two skeletons in space is never shown as an error.

No. Turning to move in the source's directions and walking where the character faces are both right; RetargetProof reads which one was made from the whole skeleton, or from the hips' own path when ordinary errors blur it, and says so in a note. When neither tells, it says that too and leaves the hips' travel alone. Bones turned one way while the hips travel the other way are reported.

Yes. It measures the bones the mesh follows and, for a fix, finds the keyed control that turns each of them. A generated Rigify rig was checked in FK and in IK. Where a limb is held by an IK control, it says so and offers no fix for that limb.

A copy of the target's action; the original action stays in the file. The source is never changed. After each fix the same check runs again and the panel shows before and after. A fix the check does not confirm is taken back.

Fingers, faces, tails, four-legged rigs, hand-to-hand and hand-to-prop contacts. "Arm inside the body" is an estimate from the skeleton, not from the mesh. A time shift between source and target is reported as not explained. A bone that changes length (a scale key) is reported, not fixed.

No.

Yes. The Whole Delivery panel checks every file in a folder you pick, each with the Source and Target stored in its scene, and writes one report. A file with none chosen is reported as not checked, never as passed.