Does Rive require characters to be positioned in a T-pose when rigged (I assume not), and does it require that each object (body, arms, legs, head, eyes, etc) within a character be a separate object/layer/group within the SVG document?
I suspect many existing SVGs will be flattened/simplified/optimized when exported (from Illustrator, Inkscape, etc) which make make it difficult to extrapolate the various parts for animation.
I’ve included a quick semi-structured/grouped sample SVG file along with the original Affinity Designer document (attached) to experiment. So far it generally seems to come into Rive (via Affinity) just fine, except for rounded stroke end-caps and joints (rendered a square) which likely means all lines/strokes must be expanded to curves (see: Bean Outlines.svg in the archive) before importing into Rive to ensure the geometry remains intact.
Bean.zip (28.2 KB)