Not every support region is worth the same amount of worry
A support touching a hidden interior wall with easy access is a non-event; the same support touching a visible face next to a thin, breakable detail with almost no room to reach it is a real risk. Treating every support region as equally concerning either causes wasted redesign effort on regions that do not matter, or blind spots on the ones that do. This tool classifies a single region at a time by the three factors that actually drive real-world risk: visibility, nearby fragility, and how hard it is to physically reach.
How the classification works
Each factor you check adds to a risk score: a visible contact surface, a fragile feature nearby, and tight or enclosed removal access. A high combined score recommends redesigning the geometry to avoid needing support here at all, since managing a high-risk removal carefully is often more effort than avoiding the need in the first place. A moderate score suggests support-setting adjustments (a lighter interface density) rather than a full redesign; a low score needs no special handling.
Worked example
A visible contact surface, a fragile feature nearby, tight enclosed access: all three risk factors are present, landing in the high band, so the tool recommends redesigning the geometry (a chamfer, a repositioned feature, or reorientation) rather than attempting a careful but genuinely difficult removal in a tight space next to something breakable.
A common mistake
A frequent error is classifying a support region's risk once at the design stage and never reconsidering it after switching materials. A more brittle material can turn a moderate-risk removal into a high-risk one, since a technique that flexed harmlessly on one material can crack a more brittle one at the same contact point. Re-check risk after any material change.
Limitations
This classifies support-contact risk from the factors you describe; it does not know your specific slicer's support settings or your own hand skill at cleanup, both of which meaningfully affect the real outcome for a moderate-risk region.