Appendix A. Known Limitations Register

Carried from the research report. Each limitation is stated with its operational consequence and the practical mitigation available.

Table 113. Known Limitations Register

#	Limitation	Operational consequence	Mitigation available
1	The framework has never been deployed in a real organization	Every threshold and cadence is a reasoned default, not a calibrated one	Instrument from day one (Chapter 29)
2	No systematic academic literature review was performed	Claims are reasoned, not literature-benchmarked	Treat design rationale as argument, not evidence
3	Risk-tier self-scoring remains gameable in the mid-range	Under-tiering is likely and the safety floor will not catch it	Countersignature and biased audit sampling (Chapter 7)
4	The two-body model does not scale on volume within one jurisdiction	Bodies become bottlenecks as adoption succeeds	Pattern expansion, delegation by decision type (Chapter 20, §20.6)
5	No cost, staffing, or service-level model exists	You cannot size the function from the framework	Measure your own; Chapter 29
6	The Core Model’s minimality is asserted, not derived	The eight entities may be wrong in composition	Record cases the model cannot express
7	The maturity model is asserted, not observed	Progression may not work as described	Use it as a checklist, not a prediction
8	The NIST and ISO crosswalk is reasoned, not clause-verified	Do not rely on it for a compliance claim	Verify independently before external use
9	Individual due-process and contestability are a declared non-goal	The framework provides an interface point, not a mechanism	Address through your regulatory overlay and legal function
10	The narrowed autonomy rule and audit sampling have not been tested together	Boundary-widening-only classification may under-catch	Weight sampling toward agentic systems (Chapter 17, §17.4)
11	No mechanism exists for cumulative or structuring exposure	Sub-threshold agent actions can aggregate to material exposure	Add aggregate limits outside the framework (Chapter 13, §13.4)
12	Cosmetic versus substantive change has no bright-line test	Classification drifts toward the cheaper option	Second opinion above a tier; retrospective sampling (Chapter 14, §14.3)
13	Security Architect is consulted, never accountable	Security under-weighted in agent failure modes	Elevate to co-accountable locally (Chapter 21, §21.4)
14	Gate consolidation reduced labels more than burden	Do not promise a lighter process	Pursue burden reduction through evidence-as-code (Chapter 4, §4.4)