Backfill Roux clean-room hierarchy

docs/HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md

@@ -1295,6 +1295,11 @@ Current status:
 - `implemented_live_clean_room_verified`
 - preserve as the earliest landed restrictive clean-room precedent in this
   audit snapshot
+- use
+  [HYPERTWIST_ROUX_METHOD_STAGE_CLEAN_ROOM_HIERARCHY_RECONCILIATION_2026-05-27.md](C:/HyperTwist/docs/HYPERTWIST_ROUX_METHOD_STAGE_CLEAN_ROOM_HIERARCHY_RECONCILIATION_2026-05-27.md:1)
+  for the current owner split across first-party training/coaching, the
+  permissive trainer-foundation lanes, and this row's bounded landed
+  method-stage clean-room family
 
 Licensing posture:
 
@@ -1336,6 +1341,10 @@ Future widening rule:
 - widen only from first-party Unreal outputs and scrubbed clean-room specs
 - keep the Model A / Model B lineage explicit
 - never treat the restrictive mirror as implementation-ready source
+- do not reinterpret the current first-party training-session, review-plan,
+  queue, follow-up, or coaching orchestration family through this row
+- do not treat this landed precedent as permission to skip clean-room
+  governance for later restrictive rows
 
 ## Bottom line
 

docs/HYPERTWIST_PHASE_1R_RETAINED_SET_CONTRACT_AND_HANDOFF_2026-05-13.md

@@ -232,6 +232,11 @@ Contract:
 - keep the Model A / Model B lineage explicit
 - read [HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md](C:/HyperTwist/docs/HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md:1) and [HYPERTWIST_MODEL_A_MAX_VALUE_EXTRACTION_MODUS_OPERANDI_2026-05-13.md](C:/HyperTwist/docs/HYPERTWIST_MODEL_A_MAX_VALUE_EXTRACTION_MODUS_OPERANDI_2026-05-13.md:1) before widening this lane
 - never treat the donor mirror itself as implementation-ready source
+- `2026-05-27` Roux hierarchy clarification: this row keeps only the landed
+  bounded method-stage, blockbuilding, and generated stage-trainer family
+  already materialized in first-party outputs; it does not reopen the broader
+  first-party training/coaching runtime and it does not relax restrictive
+  governance for later rows
 
 ## Active implementation-board rows
 

docs/HYPERTWIST_ROUX_METHOD_STAGE_CLEAN_ROOM_HIERARCHY_RECONCILIATION_2026-05-27.md

@@ -0,0 +1,97 @@
+# HyperTwist Roux Method-Stage Clean-Room Hierarchy Reconciliation - 2026-05-27
+
+## Status
+
+This document is the source-backed reconciliation note for the landed
+restrictive clean-room `Roux` lane:
+
+- `onionhoney/roux-trainers`
+
+It answers one narrow question:
+
+- does the current landed `Roux` clean-room precedent require reopening the
+  already landed first-party training/coaching implementation or broadening the
+  row into a general trainer-foundation owner
+
+Result:
+
+- no product-code reopening is required
+- yes selective authority backfill is required so the feature canon keeps the
+  landed `Roux` method-stage family visible without weakening restrictive
+  governance
+
+## Safe source basis
+
+- `C:\HyperTwist\docs\HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md`
+- `C:\HyperTwist\docs\HYPERTWIST_PHASE_1R_RETAINED_SET_CONTRACT_AND_HANDOFF_2026-05-13.md`
+- `C:\HyperTwist\docs\REPO_LICENSE_TRACKING.md`
+- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\FEATURE_REGISTRY.md`
+- `C:\Workspaces\HyperTwist\repos.manifest.json`
+- `C:\Workspaces\HyperTwist\clean-room-specs\onionhoney-roux-trainers.model-a.md`
+- `C:\Workspaces\HyperTwist\clean-room-specs\onionhoney-roux-trainers.model-b.md`
+- first-party landed-state surfaces:
+  - `C:\HyperTwist\UnrealHyperTwist\Source\UnrealHyperTwist\Private\HyperTwistTraining\HyperTwistTrainingCatalogLibrary.cpp`
+  - `C:\HyperTwist\UnrealHyperTwist\Content\HyperTwistTraining\MaterializedCatalog\hypertwist-content-pack-manifests.json`
+  - `C:\HyperTwist\UnrealHyperTwist\Content\HyperTwistTraining\MaterializedCatalog\hypertwist-runtime-content-bundles.json`
+  - `C:\HyperTwist\UnrealHyperTwist\Content\HyperTwistTraining\MaterializedCatalog\raw-extracts\restrictive-static-exercises.jsonl`
+  - `C:\HyperTwist\UnrealHyperTwist\Content\HyperTwistTraining\MaterializedCatalog\raw-extracts\restrictive-generative-exercise-surfaces.jsonl`
+
+## Slice-local verdict
+
+| Slice | Primary authority | Secondary value kept | Route | Width | Result |
+|---|---|---|---|---|---|
+| Landed first-party training-session runtime, review-plan state, follow-up queues, coaching orchestration, and broader trainer-shell integration | first-party training and coaching surfaces | the landed restrictive Roux lane remains a subordinate bounded method-stage family | existing landed first-party route | `F2` | unchanged live implementation owner |
+| Broad algorithm-training shell, set/subset corpus, and smartcube-capable drill foundation | `tao-yu/Alg-Trainer` | `Lykos/cube_trainer`, the permissive micro-drill rows, and the restrictive Roux lane remain narrower adjacent owners | existing landed permissive route | `F2` | unchanged landed trainer-foundation owner |
+| Roux method-stage drill flows, blockbuilding trainers, static case decks, and generated random-state stage surfaces | landed `onionhoney/roux-trainers` clean-room precedent | first-party training shell remains the live orchestration owner above the clean-room outputs | existing landed clean-room route | `F2` | unchanged landed restrictive method-stage owner |
+
+## Why no product-code reopening is required
+
+The current first-party landed surfaces already materialize the bounded family
+that HyperTwist meant to keep from this lane:
+
+- Roux and blockbuilding content-pack structure
+- static case decks
+- generated stage-trainer surfaces
+- first-party wording and scaffolding above the clean-room outputs
+
+That is already the correct narrow result. It does not reopen the broader
+training foundation, the first-party training-session runtime, or the default
+future restrictive queue.
+
+The governance point matters too:
+
+- this row remains the earliest landed restrictive clean-room precedent
+- that precedent does not make later restrictive mirrors implementation-ready
+- it also does not authorize direct donor reuse or Model B source access
+
+## Selective backfill result
+
+### Earlier preserve note reopened
+
+- `C:\HyperTwist\docs\HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md`
+
+### Earlier retained-set contract reopened
+
+- `C:\HyperTwist\docs\HYPERTWIST_PHASE_1R_RETAINED_SET_CONTRACT_AND_HANDOFF_2026-05-13.md`
+
+### Legal tracker reopened
+
+- `C:\HyperTwist\docs\REPO_LICENSE_TRACKING.md`
+
+### Feature canon reopened
+
+- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\FEATURE_REGISTRY.md`
+
+### Manifest notes reopened
+
+- `C:\Workspaces\HyperTwist\repos.manifest.json`
+
+## Final call
+
+`onionhoney/roux-trainers` stays the landed restrictive clean-room owner only
+for the exact bounded `Roux` method-stage and blockbuilding training slice now
+materialized in first-party outputs.
+
+It does not become a broad trainer-foundation owner, it does not reopen the
+already landed first-party training/coaching runtime, and it does not weaken
+restrictive governance for later rows.

docs/REPO_LICENSE_TRACKING.md

@@ -917,6 +917,7 @@ Source basis:
 - `C:\Workspaces\HyperTwist\clean-room-specs\onionhoney-roux-trainers.model-a.md`
 - `C:\Workspaces\HyperTwist\clean-room-specs\onionhoney-roux-trainers.model-b.md`
 - `C:\HyperTwist\docs\HYPERTWIST_LIVE_LANES_SOURCE_AND_PRESERVATION_AUDIT_2026-05-13.md`
+- `C:\HyperTwist\docs\HYPERTWIST_ROUX_METHOD_STAGE_CLEAN_ROOM_HIERARCHY_RECONCILIATION_2026-05-27.md`
 
 Practical obligations:
 
@@ -928,6 +929,8 @@ Practical obligations:
 Approved working posture:
 
 - preserve the repo as a landed restrictive clean-room precedent only
+- preserve the landed bounded Roux method-stage and blockbuilding family as a
+  first-party output family above the clean-room lineage
 - do not treat the existing clean-room implementation work as permission for later direct source reuse
 - start future widening from first-party outputs and scrubbed clean-room specs only
 

docs/v6_5_deep_manual_pack/HyperTwist/FEATURE_REGISTRY.md

@@ -186,6 +186,7 @@ repo.
 | Persisted coaching, weighted sampling, and BLD-domain session modeling | Implemented now | landed `Lykos/cube_trainer` preserve lane + first-party training stack | Current live persistence-heavy coaching adjunct remains grounded in the landed permissive `cube_trainer` preserve lane for weighted sampling, method exploration, and BLD-domain training-session logic. This does not displace the broader `Alg-Trainer` algorithm-shell/case-corpus foundation, the `KubeTimr` timer substrate, or the current first-party trainer-shell and follow-up orchestration union. |
 | Cross-planning scramble ladder and CFOP cross micro-drill progression | Implemented now | landed `newyork-anthonyng/rubiks-cross-trainer` preserve lane + first-party training stack | Current live cross-planning track remains grounded in the landed permissive `rubiks-cross-trainer` row for its explicit `1`-move through `8`-move scramble ladder and narrow cross-drill corpus. This does not displace the broader `Alg-Trainer` foundation, the `cube_trainer` persistence-heavy coaching slice, or the current first-party trainer-shell and review-plan orchestration union. |
 | Advanced `5-style` edge-cycle and blindfold micro-drill progression | Implemented now | landed `abunickabhi/5style-Trainer` preserve lane + first-party training stack | Current live advanced `5-style` / BLD micro-drill track remains grounded in the landed permissive `5style-Trainer` row for narrow mover-subset drilling, timed reveal flow, and high-skill edge-cycle practice. This does not displace the broader `Alg-Trainer` foundation, the `cube_trainer` persistence-heavy coaching slice, or the current first-party trainer-shell and follow-up orchestration union. |
+| Roux method-stage, blockbuilding, and generated stage-trainer family | Implemented now | landed `onionhoney/roux-trainers` clean-room preserve lane + first-party training stack | Current live Roux and blockbuilding family remains grounded in the landed restrictive clean-room `roux-trainers` row for static case decks, generated stage-trainer surfaces, and method-stage drill structure. This does not displace the broader `Alg-Trainer` foundation, the `cube_trainer` persistence-heavy coaching slice, or the current first-party trainer-shell and review-plan orchestration union, and it does not relax restrictive governance for later rows. |
 | Guided beginner lesson-state and stepwise pedagogy benchmark | Deep-source grounded retained | `AviKaufman/Rubix-cube-trainer` retained clean-room benchmark lane | Retained as `A1 + R4 + F2` only for named beginner milestones, oversatisfied-step explanation, bounded step help, and manual-turn/tutorial coherence on a live cube surface. This does not displace the first-party training-session, method-drill, coaching queue/follow-up, or bounded classic-cube `Phase 6R-Q/R/S` explanation and correction seams. |
 | Guided task-planner and focus-playback benchmark | Deep-source grounded retained | `alinen/cube` retained clean-room benchmark lane | Retained as `A1 + R4 + F2` only for a separable lesson-task graph, bounded task-local planner, focus-target cueing, and queued move demonstration on a visible cube surface. This does not displace the first-party training-session, method-drill, coaching queue/follow-up, canonical replay, or bounded classic-cube `Phase 6R-Q/R/S` explanation and correction seams. |
 | Blindfold memo attempt and validation benchmark | Deep-source grounded retained | `ambisinister/blindsolve` retained clean-room benchmark lane | Retained as `A1 + R4 + F2` only for a narrow blindfold memo attempt-phase loop, conceal/reveal validation, letter-pair token entry, parity-aware memo expectation, and lightweight repeat cadence. This does not displace the first-party timing-policy templates, coaching-memory/follow-up surfaces, replay-verification blindfold event taxonomy, or the broader memory doctrine. |