These drafts turn the checklist into pattern applications, interfaces and acceptance obligations. They are not completion certificates. WM-01–05 were developed first and WM-02 includes subsequent P1 design decisions; WM-06–14 now extend the same design coverage. Build coordination is now with claude-3 under individually bounded WM-only instructions. Silence between bellbacks is not evidence that the specifications have been validated.
| Group | What it must supply or constrain |
|---|---|
| WM-01–05 | Sound carriers and actual belief, dynamics, observation and prediction inputs. |
| WM-06 and WM-13 | A compatible preference member, its family, composition and revision authority. |
| WM-08, WM-09 and WM-14 | Real source-bound patterns, attributed candidate construction and in-loop interpretation/search improvement. |
| WM-10 and WM-11 | Canonical cascade G, strict improvement admission for extensions, applicable policy inference and actual choice. |
| WM-12 | Comparable observation opportunity, hierarchy, guards/history and effective anticipation configuration. |
| WM-07 | Actual parameter inference and its separately justified epistemic consumer. |
These groups constrain one another. They are not nine serial prerequisites for finishing the current P1 packet. The build lead may execute only the explicitly dispatched change and must return scope questions before expanding it. Design findings return through bells and are retained as pattern applications or revisions.
All WM-01–14 now have authored cascades. The R-node, TRACE and E01–E10 obligations now have 32 separate design cascades. They remain unapproved for implementation until explicitly greenlit. Their relevant evidence constraints must be retained in build packets from the beginning. This publication neither changes checklist checkboxes nor claims the current deployed state has been re-audited.
The current cascade semantics source requires G for partial constructions, per-problem candidate generation and strict delta-G improvement for admitting extensions. It supersedes the old global formula-fork framing. Remaining episode, fuel, board-transition and yield proposals keep their stated unresolved status. Further rulings can revise these designs without rewriting historical evidence.