Why we tried it
Express custody coverage and combinatorial constraints as checkable deterministic witnesses.
What the evidence shows
Synthetic witnesses validate the supplied instance, not the deployed protocol distribution or threat model.
Where the work stands
Protocol mappings, correlations and adversary capabilities need actual modeling; synthetic coverage is not a live security probability.
Possible next steps
Pin the protocol version and threat model, audit the incidence-model mapping, replay the witnesses independently and test mismatch cases.
Suggested next steps are untested research questions, not new experimental results. The archive records manuscript provenance, licensing and status.