[irl-dan]

The authored periodic-review reference listed monitoring values without their metric names or inclusive threshold boundaries, and omitted the exact metric identity from issue descriptions. This obscures what the numbers measure and falls short of the existing house policy's scope requirements. The previous checker passed because it checked synthetic rule calculations rather than these report bindings.

Add all three metric names, available observation dates, inclusive limits and both exact issue scopes. Clarify that the 0.03 observation is a calibration error in probability units, not a predicted default probability. The reference remains 650 words excluding its identity appendix, below the existing 850-word limit. Extend only fixed-reference source correspondences and eight targeted mutations; this is not a general prose evaluator.

All 35 offline source commands pass, including twenty inherited unit tests. All 455 other original files and all 32 component-package files remain identical. No agreement, input packet, numerical observation, runtime behavior or reusable definition changes. No financial or provider calls occurred; reporting and assessment remain agent-unqualified.

Draft stacked on #39. No merge or release requested.