Some risks are reversible if they materialise and some are not. Any system that treats the two categories as points on the same scale will occasionally recommend a defensible response to a catastrophic risk, and defensibility after the fact is not the standard the reader is looking for.
A family office weighing a counterparty exposure that is unlikely but would be terminal if it materialised is not primarily interested in the probability. It is interested in whether the exposure can be closed at any cost that is small relative to the terminal outcome, and whether the closure can be made durable without the counterparty becoming aware of it.
The mechanics require an explicit category for the terminal case. Every finding is tagged for reversibility at the time it is recorded. Terminal findings are routed on a separate track with a distinct response protocol and act one confidence band earlier than the general framework would require. The rationale is written into the framework rather than left to the reader's discretion, so a rotation does not silently downgrade the response.
Reversible risks are managed by expected value. Terminal ones are managed by exclusion. Confusing the two is a design decision the reader is entitled to reject in advance.