diff --git a/docs/REPO_LICENSE_TRACKING.md b/docs/REPO_LICENSE_TRACKING.md index 9d56b1e..3a6475e 100644 --- a/docs/REPO_LICENSE_TRACKING.md +++ b/docs/REPO_LICENSE_TRACKING.md @@ -155,8 +155,10 @@ Status update on `2026-05-21`: packet is now landed in current code - the bounded permissive `Phase 6R-Q` `vivaansinghvi07/rubix-cube-solver` solve explanation/recommendation packet is now landed in current code -- the current next bounded move is a source-backed retained classic-cube recognition multi-face - correction/explanation preparation/control pass, not a new restrictive packet by default +- the generic source-backed `Phase 6R-R` shared classic-cube recognition multi-face + correction/explanation control pass is now consumed +- the current next bounded move is the bounded permissive `Phase 6R-R` classic-cube recognition + multi-face correction/explanation shell packet, not a new restrictive packet by default - use the repo-row README census, the portfolio standing refresh backfill, and the `2R-A` ownership contract for the current queue after that correction @@ -260,8 +262,8 @@ Current practical interpretation: - the landed `PostHog/posthog` boundary-sensitive widening packet now lives in `docs/HYPERTWIST_PHASE_4R_PACKET_4R_D_POSTHOG_CONTROL_PLANE_IMPLEMENTATION_2026-05-13.md` - the landed `screenpipe/screenpipe` boundary-sensitive widening packet now lives in `docs/HYPERTWIST_PHASE_4R_PACKET_4R_E_SCREENPIPE_CAPTURE_HISTORY_IMPLEMENTATION_2026-05-13.md` - the landed `remotion-dev/remotion` boundary-sensitive widening packet now lives in `docs/HYPERTWIST_PHASE_4R_PACKET_4R_F_REMOTION_MEDIA_EXPORT_IMPLEMENTATION_2026-05-13.md` -- the next bounded move is a source-backed retained classic-cube recognition multi-face - correction/explanation preparation/control pass +- the next bounded move is the bounded permissive `Phase 6R-R` classic-cube recognition + multi-face correction/explanation shell packet Companion docs: @@ -752,7 +754,8 @@ Approved working posture: - classic-cube webcam shell - keep the following families deferred to later queue decisions: - bundled-font redistribution and multilingual solve shell - - multi-face reconstruction / correction / explanation + - direct standalone multi-face reconstruction / correction / explanation + - only the shared `Phase 6R-R` capture-state adjunct remains live now - if later implementation copies bundled non-code assets, capture and preserve any asset-specific license or replace the asset with a clearly redistributable alternative @@ -805,7 +808,8 @@ Approved working posture: - browser/webcam shell profile and browser-shell session-state composition - bounded solve explanation/recommendation stage ladder and recommendation state - keep the row only partially incorporated by default: - - multi-face correction or explanation shell beyond the bounded stage ladder remains deferred + - multi-face correction or explanation shell beyond the bounded stage ladder is the queued + `Phase 6R-R` implementation slice - any redistribution of `frontend/lib/twistysim.min.js` still requires preserved or replaced upstream provenance/license diff --git a/docs/arch/HYPERTWIST_PHASE6R_R_CLASSIC_CUBE_MULTI_FACE_CORRECTION_EXPLANATION_PREPARATION_PACKET_2026-05-23.md b/docs/arch/HYPERTWIST_PHASE6R_R_CLASSIC_CUBE_MULTI_FACE_CORRECTION_EXPLANATION_PREPARATION_PACKET_2026-05-23.md new file mode 100644 index 0000000..d76165b --- /dev/null +++ b/docs/arch/HYPERTWIST_PHASE6R_R_CLASSIC_CUBE_MULTI_FACE_CORRECTION_EXPLANATION_PREPARATION_PACKET_2026-05-23.md @@ -0,0 +1,262 @@ +# HyperTwist Phase 6R-R classic-cube multi-face correction explanation preparation packet + +Created on `2026-05-23` + +## Status + +- historical consumed preparation authority +- bounded post-`Phase 6R-Q` control slice +- the first bounded implementation slice now lands separately after this packet + +## Purpose + +This packet freezes the next widening order after the landed `Phase 6R-Q` +`vivaansinghvi07/rubix-cube-solver` solve explanation/recommendation slice. + +The open task was: + +- define the next bounded source-backed widening packet as the shared + classic-cube recognition multi-face correction/explanation shell above the + already landed `qbr` webcam shell and `rubix-cube-solver` reconstruction, + browser-shell, and bounded recommendation seams + +It is not: + +- a full `qbr` row transplant +- a full `rubix-cube-solver` row transplant +- a bundled `twistysim.min.js` redistribution packet +- a bundled `arial-unicode-ms.ttf` redistribution packet +- a multilingual `qbr` shell packet +- a generalized solver backend packet +- a broad playback-runtime packet + +## Current authority basis + +This preparation packet stands on already-closed authority: + +- `docs/REPO_LICENSE_TRACKING.md` +- `docs/v6_5_deep_manual_pack/HyperTwist/ROADMAP.md` +- `docs/arch/HYPERTWIST_PHASE6R_O_QBR_WEBCAM_UI_SHELL_IMPLEMENTATION_PACKET_2026-05-22.md` +- `docs/arch/HYPERTWIST_PHASE6R_P_RUBIX_CUBE_SOLVER_BROWSER_WEBCAM_SHELL_IMPLEMENTATION_PACKET_2026-05-22.md` +- `docs/arch/HYPERTWIST_PHASE6R_Q_RUBIX_CUBE_SOLVER_SOLVE_EXPLANATION_RECOMMENDATION_IMPLEMENTATION_PACKET_2026-05-23.md` + +The key accepted routing facts are: + +- `kkoomen/qbr` remains the primary classic-cube live-recognition and + capture-state donor +- `vivaansinghvi07/rubix-cube-solver` remains the stronger correction, + reconstruction-companion, and explanation-semantics donor +- the already landed `Phase 6R-O` slice owns: + - classic-cube webcam shell profile + - scanned-side and captured-face shell state + - target-face hint and bounded panel visibility +- the already landed `Phase 6R-P` slice owns: + - browser-shell session-state composition + - browser-stage routing + - websocket transport metadata + - manual cube-edit handoff +- the already landed `Phase 6R-Q` slice owns: + - bounded solve explanation stage ladder + - recommendation-state readout + - explanation readiness after full reconstruction + +The donor value now strongest for the next narrower slice is: + +- multi-face correction-target state above the landed capture and + reconstruction seams +- contradiction-aware correction and re-scan guidance +- correction explanation readout for incomplete, conflicting, or manually + revised cube state +- explicit correction-target routing that composes `qbr` target-face capture + state with `rubix-cube-solver` correction/explanation semantics + +Ownership denied here is: + +- do not widen into bundled `twistysim.min.js` redistribution +- do not widen into bundled `arial-unicode-ms.ttf` redistribution +- do not widen into standalone multilingual `qbr` ownership +- do not widen into generalized solver backend ownership +- do not widen into broad playback-runtime ownership + +## Cross-lane retained split + +Exact contested slice: + +- classic-cube recognition multi-face correction/explanation shell + +Candidate repos considered: + +- `kkoomen/qbr` +- `vivaansinghvi07/rubix-cube-solver` + +Authority outcome: + +- `vivaansinghvi07/rubix-cube-solver` - `A1 / R1 / F2` + - retained primary owner for correction/explanation semantics, manual + correction flow, and structured multi-face correction readout +- `kkoomen/qbr` - `A2 / R1 / F2` + - retained adjunct owner for capture-state continuity, target-face re-scan + anchoring, preview/snapshot/result-state discipline, and shell-state + continuity needed by the shared correction surface + +The top-level product shell owner does not change here: + +- first-party HyperTwist remains the broader recognition-session and + product-shell owner + +## Why this was the next packet + +The earlier retained queue heads were already landed: + +- `Phase 6R-A` `HactarCE/Hyperspeedcube` +- `Phase 6R-B` `kkoomen/qbr` +- `Phase 6R-C` `vivaansinghvi07/rubix-cube-solver` +- `Phase 6R-D` `roice3/MagicTile` +- `Phase 6R-E` `ggml-org/whisper.cpp` +- `Phase 6R-F` `SYSTRAN/faster-whisper` +- `Phase 6R-G` `rhasspy/piper` +- `Phase 6R-H` `coqui-ai/TTS` +- `Phase 6R-I` `roice3/Magic120Cell` +- `Phase 6R-J` `roice3/MagicCube5D` +- `Phase 6R-K` `HactarCE/Hyperspeedcube` +- `Phase 6R-L` `HactarCE/Hyperspeedcube` +- `Phase 6R-M` `HactarCE/Hyperspeedcube` +- `Phase 6R-N` `HactarCE/Hyperspeedcube` +- `Phase 6R-O` `kkoomen/qbr` +- `Phase 6R-P` `vivaansinghvi07/rubix-cube-solver` +- `Phase 6R-Q` `vivaansinghvi07/rubix-cube-solver` + +That advanced the live queue to: + +- shared `Phase 6R-R` classic-cube recognition multi-face correction/ + explanation + +## Required result + +The source-backed control pass for this packet is now complete. + +The first actual `6R-R` implementation packet should: + +- define one bounded shared correction/explanation shell slice +- keep the slice above the landed `Phase 6R-O`, `Phase 6R-P`, and `Phase 6R-Q` + seams +- widen only the first correction-state and correction-explanation family that + can stand on its own without dragging in either bundled asset family +- preserve `rubix-cube-solver` as the retained correction/explanation-semantic + owner while preserving `qbr` as the capture-state adjunct +- explicitly state which neighboring retained rows stay closed in that packet + +## Source-backed retained basis + +The queue-head decision is now source-backed rather than README-only. + +Inspected retained donor basis: + +- `C:\visual_studio_solutions\multi_project\GPT 5.4 HyperTwist parse\19-kkoomen-qbr-upstream-dossier.md` + with code-level findings from: + - `README.md` + - `src/video.py` + - `src/helpers.py` + - `src/colordetection.py` + - `src/config.py` + - `src/constants.py` +- `C:\visual_studio_solutions\multi_project\GPT 5.4 HyperTwist parse\20-vivaansinghvi07-rubix-cube-solver-upstream-dossier.md` + with code-level findings from: + - `backend/cv.py` + - `backend/server.py` + - `frontend/src/script.js` + - `frontend/lib/twistysim.min.js` + +Inspected first-party receiving basis: + +- `UnrealHyperTwist/Source/UnrealHyperTwist/Public/HyperTwistRecognition/HyperTwistRecognitionTypes.h` +- `UnrealHyperTwist/Source/UnrealHyperTwist/Private/HyperTwistBootstrap/HyperTwistContractLibrary.cpp` +- `UnrealHyperTwist/Source/UnrealHyperTwist/Private/HyperTwistTraining/HyperTwistTrainingSubsystem.cpp` +- `UnrealHyperTwist/Source/UnrealHyperTwist/Tests/HyperTwistQbrPhase6ROWebcamShellContractTest.cpp` +- `UnrealHyperTwist/Source/UnrealHyperTwist/Tests/HyperTwistRubixCubeSolverPhase6RPBrowserShellContractTest.cpp` +- `UnrealHyperTwist/Source/UnrealHyperTwist/Tests/HyperTwistRubixCubeSolverPhase6RQSolveExplanationContractTest.cpp` + +## Narrowed 6R-R slice decision + +The first widening slice is fixed as: + +1. classic-cube correction state and correction-explanation shell composition + +not as bundled asset redistribution or broader playback ownership. + +That narrowed slice covers: + +- first-party correction-target state for: + - pending captured faces + - conflicting or low-confidence committed faces + - next recommended re-scan or manual-correction target + - correction-complete versus solve-ready gating +- first-party correction-explanation readout for: + - why a correction is requested + - which face or revision is affected + - whether re-scan, manual edit, or stage restart is the correct next action + - contradiction visibility rather than a boolean-only mismatch light +- shared shell composition that keeps: + - `qbr` target-face and capture-state continuity + - `rubix-cube-solver` correction and explanation semantics + - first-party HyperTwist ownership of the wider recognition session and + shell +- subsystem normalization and focused automation coverage + +## Deferred neighboring families + +The first `6R-R` implementation packet must keep these capability families +closed: + +- bundled `twistysim.min.js` shipping or redistribution +- bundled `arial-unicode-ms.ttf` shipping or redistribution +- standalone multilingual `qbr` shell ownership +- generalized solver backend ownership +- broad playback-runtime ownership +- any donor-shaped frontend transplant + +## Why this slice was first + +- `rubix-cube-solver` is strongest for explicit correction and explanation + semantics that sit above the already landed reconstruction, browser-shell, + and recommendation seams +- `qbr` remains uniquely valuable for stable capture-state continuity and + target-face re-scan anchoring, so that adjunct value must stay preserved in + the shared shell rather than being flattened away +- current first-party code already owns the session, shell, reconstruction, + and bounded recommendation envelopes, but it does not yet own a first-party + correction-target and correction-explanation surface above them +- widening into either bundled asset family or broad playback/runtime + ownership would prematurely absorb legal and routing concerns that are still + narrower and separately governed + +## Out of scope for the first 6R-R packet + +- bundled `twistysim.min.js` shipping +- bundled `arial-unicode-ms.ttf` shipping +- standalone multilingual shell ownership +- generalized solver backend ownership +- broad playback runtime ownership +- any attempt to demote `qbr` capture-state adjunct value or to make the + retained split symmetric + +## Acceptance criteria + +- the packet records why the live queue advanced to shared `6R-R` +- the packet records that the first `6R-R` widening slice is correction state + and correction-explanation shell composition only +- the packet states the exact `A/R/F` split between `rubix-cube-solver` and + `qbr` +- the packet states exact out-of-scope families for the first `6R-R` pass +- the packet keeps both bundled-asset guards visible + +## Validation checklist + +1. confirm the shared retained correction/explanation seam narrows cleanly to + a first bounded correction-state and correction-explanation shell +2. confirm the implementation packet is framed as bounded first-party widening + rather than a donor frontend transplant +3. confirm both bundled-asset redistribution families remain explicitly + deferred + +That is the packet. diff --git a/docs/ops/HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md b/docs/ops/HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md new file mode 100644 index 0000000..87b5ab8 --- /dev/null +++ b/docs/ops/HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md @@ -0,0 +1,435 @@ +# HyperTwist Continuity Lattice And Context Assembly Profile Doctrine - 2026-05-23 + +## Purpose + +This doctrine integrates the useful context-assembly and transmission-profile +improvements from the newer retained continuity framing into HyperTwist +without replacing HyperTwist's stronger existing memory doctrine. + +It answers these questions: + +1. what `Continuity Lattice` means inside HyperTwist +2. how it relates to the canonical `Memory Lanes` +3. what `Max-Retention Mode` and `Economic-Retention Mode` mean +4. which imported ideas are accepted, and which are rejected + +## Governing relationship + +The existing memory doctrine remains canonical: + +- `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` + +That doctrine still owns: + +- lane taxonomy +- lane authority +- write and derivation rules +- provenance rules +- custody visibility +- clean-room boundaries +- implementation order + +This doctrine is additive only. + +It adds: + +- a short substrate name +- context-assembly profile terminology +- transmission and retrieval posture +- explicit override scope +- escalation and cost-aware routing rules + +It does not replace or demote: + +- lane authority +- provenance law +- custody law +- clean-room boundaries +- raw-authority preservation +- contradiction visibility +- non-regression rules + +Do not treat this as a second memory system. + +Do not treat this as a replacement memory theory. + +## Canonical taxonomy + +Use these short names inside the product taxonomy: + +1. `Continuity Lattice` + - the federated continuity, memory, provenance, retrieval, and + context-assembly substrate +2. `Memory Lanes` + - the bounded lane taxonomy defined in the canonical memory doctrine +3. `Max-Retention Mode` + - the high-fidelity context-assembly profile +4. `Economic-Retention Mode` + - the authority-aware token-economical context-assembly profile + +The correct relationship is: + +- `Continuity Lattice` is the substrate name only +- `Memory Lanes` remain the canonical internal memory taxonomy +- `Max-Retention Mode` and `Economic-Retention Mode` are context-assembly + profiles over the same substrate + +This is not a second memory system. + +## What HyperTwist already kept and what this doctrine adds + +HyperTwist already had the stronger memory architecture: + +- bounded memory lanes instead of one monolithic memory blob +- raw authority surviving derivation +- session continuity distinct from cognition +- shared coordination distinct from durable memory +- optional assistive reducers and compactors + +This doctrine therefore imports only what actually improves HyperTwist: + +1. explicit substrate wording +2. explicit context-assembly profile wording +3. transmission and retrieval posture +4. explicit override scope +5. escalation law +6. cost-aware routing language + +## Adopted continuity-profile rules + +The following imported ideas fit HyperTwist and are now adopted: + +1. one substrate, two context-assembly profiles +2. storage retention is separate from active transmission +3. raw authority remains preserved in both modes +4. summaries and compact views remain derived and optional +5. delta transmission is economically stronger than naive replay +6. retrieval should be lane-aware, authority-aware, and custody-aware +7. context assembly should stay reversible enough to reopen raw authority +8. escalation should happen before uncertainty becomes dangerous + +State these rules explicitly: + +- storage retention != active transmission +- one substrate, two context-assembly profiles +- raw authority remains preserved in both modes +- economic behavior must never silently replace truth + +## Rejected flattening + +The following imported tendencies are explicitly rejected: + +1. one single authority ladder for everything +2. confidence as a substitute for authority +3. any framing that flattens restrictive custody or clean-room boundaries into + a generic relevance score +4. any framing that lets compact or derived views quietly replace raw + authority +5. any framing that lets economic optimization override provenance, custody, + non-regression, or legal boundaries + +The governing rule is: + +- confidence does not outrank authority + +## Profile definitions + +`Max-Retention Mode` and `Economic-Retention Mode` are: + +- preset-backed defaults +- overrideable +- not rigid bundles +- not separate storage systems +- not separate truth systems + +They differ in: + +- context-assembly posture +- transmission aggressiveness +- raw-authority injection behavior +- retrieval and escalation policy + +They do not differ in: + +- memory ownership +- truth rank +- custody law +- provenance law + +## Max-Retention Mode + +Purpose: + +- maximize reasoning fidelity when omission risk is more dangerous than token + cost + +Use when: + +- architecture doctrine is active +- memory governance is active +- clean-room extraction is active +- cross-repo consolidation is active +- contradiction-heavy work is active +- non-regression-sensitive work is active +- the user explicitly requests omission-intolerant handling + +Rules: + +- preserve rich authority injection +- prefer raw authority near the model +- use derived summaries only as navigation aids +- keep contradiction state explicit +- keep donor and clean-room boundary state explicit + +## Economic-Retention Mode + +Purpose: + +- maximize useful capability per transmitted token without deleting truth + +Use when: + +- routine coding is active +- local debugging is active +- bounded feature work is active +- repeated build-fix loops are active +- known-scope implementation is active +- contradiction and boundary risk remain low + +Rules: + +- retain all authority in storage +- inject only what the current reasoning boundary requires +- prefer delta transmission +- prefer retrieval-first assembly +- prefer stable prefixes for invariant doctrine +- use summaries only as routing and navigation surfaces + +## Escalation law + +Escalate from `Economic-Retention Mode` to `Max-Retention Mode` before +uncertainty becomes dangerous. + +Do not wait for silent semantic erosion. + +Escalate if any of the following becomes true: + +1. contradiction is detected +2. clean-room or restrictive custody becomes active +3. source authority becomes unclear +4. doctrine changes or branch or `HEAD` drift invalidate a compact packet +5. stale derived-view detection fires +6. unresolved contradiction count rises above the safe bound +7. the user requests exhaustive or omission-intolerant handling + +## Preset and override law + +`Max-Retention Mode` and `Economic-Retention Mode` should be implemented as +first-party preset profiles over the same `Continuity Lattice`. + +The expected control model is: + +1. profile selection + - `Max-Retention Mode` + - `Economic-Retention Mode` + - later custom profiles where useful +2. optional assistive feature toggles + - per-feature control for compactors, token reducers, token strippers, + summarizers, memory reducers, prompt-shaping helpers, auto-triggered + skills, and similar adjuncts +3. override scope + - global + - per-project + - per-session +4. non-optional core law surface + - provenance preservation + - lane separation + - custody visibility + - clean-room boundaries + - authoritative raw-path preservation + - contradiction visibility + - rebuildability rules for indexes and derived views + +The practical rule is: + +- profiles are preset-backed defaults, not locks +- singular assistive components remain individually toggleable +- future custom profiles are allowed +- off-state behavior remains valid and supported + +Examples HyperTwist must be able to represent: + +- `Max-Retention Mode` with summarizers off +- `Economic-Retention Mode` with token stripping off +- `Economic-Retention Mode` with only delta or retrieval behavior active +- a custom profile derived from one preset but with selective overrides + +## Optional assistive adjunct governance + +Assistive adjuncts remain: + +- optional +- user-deactivatable +- removable +- non-authoritative + +This includes: + +- compactors +- token reducers +- token strippers +- summarizers +- memory reducers +- prompt-shaping helpers +- auto-triggered skills +- similar adjuncts + +The modes may choose different defaults for those components. + +They may not make those components mandatory infrastructure. + +Presets may set defaults. + +Users must still be able to override individual components. + +## Non-optional law surface + +The following are not optional assistive features and must not be toggled off +as if they were mere UX helpers: + +- provenance preservation +- lane separation +- custody visibility +- clean-room boundaries +- authoritative raw-path preservation +- contradiction visibility +- rebuildability rules for indexes and derived views + +Assistive adjuncts are optional. + +Governing safety and authority laws are not optional. + +## Architectural distinctions + +Keep these orderings separate. + +Do not collapse them into one generic importance ranking. + +### Truth arbitration order + +Use this when deciding what is true: + +1. first-party doctrine and non-regression rules +2. first-party live implementation and verified runtime state +3. lane-specific authority docs and accepted first-party packets +4. approved clean-room handoffs and deep-source donor packet conclusions +5. curated knowledge entries and user-authored notes in their own lanes +6. provenance-backed consolidated memory +7. compact or derived navigation views +8. retrieval indexes +9. shallow placeholders + +### Transmission priority + +Use this when assembling an active context packet: + +1. current explicit user instruction +2. current packet scope +3. active local diffs, files, errors, and runtime blockers +4. minimal relevant non-regression and safety constraints +5. relevant doctrine or lane slices +6. raw authority slices +7. continuity packet +8. derived navigation summaries +9. retrieval handles for dormant but reopenable authority + +Mode controls how aggressively items `5` through `9` expand. + +### Retrieval priority + +Use this when reopening stored authority: + +1. active branch and `HEAD` aligned raw records +2. same-lane authoritative records with intact provenance +3. contradiction records and unresolved conflict objects +4. custody-bound records required by the current route +5. user-authored notes and curated knowledge where relevant +6. derived navigation views +7. rebuildable indexes and stale derived packets + +### Custody visibility + +Use this when deciding what a given implementation actor may see: + +1. first-party authoritative surfaces +2. approved Model B safe handoffs +3. public or permissive reference surfaces +4. Model A restrictive source custody only where the workflow explicitly + allows it + +This is a visibility rule, not a truth-ranking rule. + +## Target-shape guidance for future implementation + +HyperTwist does not yet claim full implementation of these record shapes. + +These are target-shape requirements for later work. + +At minimum a future memory record should carry: + +- authority class +- custody class +- lane +- provenance +- stale or superseded state +- model-visibility constraints + +Future records and retrieval handles should support: + +- dormant but reopenable raw authority +- explicit model-visibility constraints +- contradiction objects rather than a boolean-only contradiction flag + +A future mode selector should consider: + +- task risk +- cost sensitivity +- clean-room boundary +- contradiction detection +- doctrine changes +- branch or `HEAD` changes +- stale derived-view detection +- unresolved contradiction count + +## Provenance and derived-memory law + +Keep these rules explicit: + +- raw capture remains authoritative +- derived views remain derived +- summaries are routing and navigation surfaces, not truth +- provenance survives every derivation +- compact views do not replace raw authority +- retrieval indexes remain rebuildable +- user-authored notes remain distinct from machine-derived memory +- continuity is not cognition +- shared coordination memory is not durable authority + +## Discovery surfaces + +This doctrine should be discoverable from: + +- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\ARCHITECTURE.md` +- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\ROADMAP.md` +- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\FEATURE_REGISTRY.md` +- `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\PROVENANCE_AND_TRUST_MODEL.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\Workspaces\HyperTwist\clean-room-specs\README.md` +- `C:\Workspaces\HyperTwist\clean-room-specs\MODEL_OUTPUT_RULES.md` +- `C:\Workspaces\HyperTwist\clean-room-specs\GENERIC_OPERATOR_SEQUENCE_AND_PROMPTING.md` +- `C:\Workspaces\HyperTwist\clean-room-specs\bounded-implementation-workflow.md` +- `C:\Workspaces\HyperTwist\clean-room-specs\0600-phase5r-clean-room-operator-sequence-and-handoff.md` + +That is the doctrine. diff --git a/docs/ops/HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md b/docs/ops/HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md index 367bea9..64a05dd 100644 --- a/docs/ops/HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md +++ b/docs/ops/HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md @@ -79,6 +79,7 @@ Use these as the current governing docs: - `C:\HyperTwist\docs\ops\HYPERTWIST_CANONICAL_COMPACT_CLOSEOUT_FORMAT_2026-05-19.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_SKILLIZATION_AND_COMMAND_SURFACE_DOCTRINE_2026-05-21.md` @@ -100,13 +101,18 @@ The next bounded move is now: packet is now landed in current code 7. the bounded permissive `Phase 6R-Q` `vivaansinghvi07/rubix-cube-solver` solve explanation/recommendation packet is now landed in current code -8. the next bounded move is a source-backed retained classic-cube recognition multi-face - correction/explanation preparation/control pass shared across `kkoomen/qbr` and - `vivaansinghvi07/rubix-cube-solver` -9. keep the `rubix-cube-solver` guard visible: +8. the continuity-profile sync now names `Continuity Lattice` as the substrate only, keeps + `Memory Lanes` canonical, and treats `Max-Retention Mode` and `Economic-Retention Mode` as + preset-backed overrideable context-assembly profiles rather than a second memory system +9. the generic source-backed `Phase 6R-R` shared classic-cube recognition multi-face + correction/explanation control pass is now consumed +10. the next bounded move is the bounded permissive `Phase 6R-R` classic-cube recognition + multi-face correction/explanation shell packet shared across `kkoomen/qbr` and + `vivaansinghvi07/rubix-cube-solver` +11. keep the `rubix-cube-solver` guard visible: - do not copy or redistribute `frontend/lib/twistysim.min.js` without preserving or replacing its upstream provenance/license in any future widening that would ship that asset -10. keep the `qbr` guard visible: +12. keep the `qbr` guard visible: - do not copy or redistribute `src/assets/arial-unicode-ms.ttf` without separate confirmation or replacement in any future widening that would ship that asset @@ -118,6 +124,18 @@ When HyperTwist later returns to memory implementation: - follow the lane map and packet order in `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +## Context-assembly sequencing rule + +When HyperTwist chooses a context-assembly posture: + +- do not replace the lane doctrine with one monolithic continuity concept +- use `Continuity Lattice` as the substrate name only +- keep `Memory Lanes` as the governing internal taxonomy +- treat `Max-Retention Mode` and `Economic-Retention Mode` as preset-backed, + overrideable context-assembly profiles over the same substrate +- escalate from `Economic-Retention Mode` to `Max-Retention Mode` before + uncertainty becomes dangerous + ## Provider-specific sequencing rule Do not implement provider-specific overlays, usage surfaces, or model-profile diff --git a/docs/ops/HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md b/docs/ops/HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md index 8ef784a..2293f82 100644 --- a/docs/ops/HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md +++ b/docs/ops/HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md @@ -65,8 +65,45 @@ Companion authorities: - `C:\HyperTwist\docs\ops\HYPERTWIST_CROSS_LANE_AUTHORITY_HIERARCHY_AND_RECONCILIATION_2026-05-20.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md` +## 2026-05-23 continuity substrate correction + +This memory doctrine remains canonical. + +HyperTwist should not replace it with a new monolithic continuity concept. + +Use `Continuity Lattice` only as the substrate short name for the federated +continuity, provenance, retrieval, and context-assembly layer that already +rests on these `Memory Lanes`. + +The required relationship is: + +- `Continuity Lattice` is the substrate name only +- `Memory Lanes` remain the governing internal taxonomy +- `Max-Retention Mode` and `Economic-Retention Mode` are preset-backed, + overrideable context-assembly profiles over the same substrate + +This import is additive only. + +It may improve: + +- context assembly +- transmission strategy +- override semantics +- explicit preset wording +- escalation logic +- cost-aware routing + +It must not demote: + +- lane authority +- provenance law +- custody law +- clean-room boundaries +- raw-authority preservation + ## Current evidence posture HyperTwist already contains real first-party memory-adjacent state in: diff --git a/docs/ops/HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md b/docs/ops/HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md index 78fea45..ef06f99 100644 --- a/docs/ops/HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md +++ b/docs/ops/HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md @@ -177,9 +177,24 @@ At minimum, support: Preferred next level: -- per-profile overrides +- per-project overrides - per-session overrides +If HyperTwist uses named context-assembly presets such as `Max-Retention Mode` +or `Economic-Retention Mode`, those presets must remain preset-backed +defaults, not rigid bundles. + +They may set defaults. + +They must remain overrideable at: + +- global scope +- per-project scope +- per-session scope + +Users must still be able to disable individual assistive adjuncts inside those +profiles. + ### Settings and menu rule If a feature is operator-visible, it should be operator-controllable. @@ -219,6 +234,59 @@ Derived state must be: - safe to discard - regenerable where practical +## Context-assembly profile interaction rule + +`Max-Retention Mode` and `Economic-Retention Mode` may choose different +defaults for optional assistive components such as: + +- compactors +- token reducers +- token strippers +- summarizers +- memory reducers +- prompt-shaping helpers +- auto-trigger skills +- similar adjuncts + +They must not make those components mandatory infrastructure. + +They must remain: + +- optional +- user-deactivatable +- removable +- non-authoritative + +The practical rule is: + +- presets may set defaults +- presets may not silently remove per-feature control +- presets may not demote the raw authoritative path +- off-state behavior remains valid and supported +- future custom profiles are allowed if they obey the same non-optional law + surface + +Examples HyperTwist must support: + +- `Max-Retention Mode` with summarizers off +- `Economic-Retention Mode` with token stripping off +- `Economic-Retention Mode` with only delta or retrieval behavior active +- a custom profile with selective assistive overrides + +## Non-optional law surface + +The following are not optional assistive features and must not be profile- +toggled off: + +- provenance retention +- authority ordering +- custody visibility +- lane separation +- clean-room boundaries +- authoritative raw-path preservation +- contradiction visibility +- rebuildability of indexes and derived views + ## Donor adoption rule Any donor-derived capability in these families must enter HyperTwist as an @@ -266,6 +334,7 @@ The acceptance standard is: Use this doctrine with: - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_SKILLIZATION_AND_COMMAND_SURFACE_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md` diff --git a/docs/v6_5_deep_manual_pack/HyperTwist/ARCHITECTURE.md b/docs/v6_5_deep_manual_pack/HyperTwist/ARCHITECTURE.md index 8117e7d..e4daa64 100644 --- a/docs/v6_5_deep_manual_pack/HyperTwist/ARCHITECTURE.md +++ b/docs/v6_5_deep_manual_pack/HyperTwist/ARCHITECTURE.md @@ -20,6 +20,10 @@ Use these alongside this document: - `WORKBENCH_FEATURE_TARGETS.md` for cockpit/workbench targets - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` for memory-lane architecture +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` + for continuity substrate and context-assembly profiles +- `C:\HyperTwist\docs\ops\HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md` + for optional assistive and off-state governance - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` for provider/BYOK architecture - `C:\HyperTwist\docs\ops\HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md` @@ -152,8 +156,37 @@ Do not treat provider-specific donors as top-level contract owners. Those two corrections are governed by: - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` +## Continuity substrate + +HyperTwist's federated continuity and memory substrate is now referred to +internally as the `Continuity Lattice`. + +That means: + +- one federated substrate +- bounded `Memory Lanes` +- explicit provenance +- explicit custody +- two context-assembly profiles: + - `Max-Retention Mode` + - `Economic-Retention Mode` +- those profiles are preset-backed defaults for context assembly, not rigid + bundles +- optional assistive components inside those profiles remain individually + controllable +- the profiles may set defaults, but they must remain overrideable and must + not weaken provenance, custody, lane separation, clean-room boundaries, or + raw-authority preservation + +It does not mean: + +- one monolithic memory blob +- summary-first authority +- a second memory system competing with the established memory-lane doctrine + ## Skillization correction Do not let donor commands or optional skills become hidden product truth. diff --git a/docs/v6_5_deep_manual_pack/HyperTwist/FEATURE_REGISTRY.md b/docs/v6_5_deep_manual_pack/HyperTwist/FEATURE_REGISTRY.md index 7dd64e0..9953d7d 100644 --- a/docs/v6_5_deep_manual_pack/HyperTwist/FEATURE_REGISTRY.md +++ b/docs/v6_5_deep_manual_pack/HyperTwist/FEATURE_REGISTRY.md @@ -47,6 +47,7 @@ Primary authority surfaces behind this registry: - `C:\HyperTwist\docs\v6_5_deep_manual_pack\HyperTwist\CAPABILITY_CONSOLIDATION_DOCTRINE.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_IMPLEMENTATION_PHASE_1_KICKOFF.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_SKILLIZATION_AND_COMMAND_SURFACE_DOCTRINE_2026-05-21.md` - the relevant implementation packet docs @@ -150,7 +151,7 @@ repo. | Bounded solve explanation/recommendation shell | Implemented now | `rubix-cube-solver` bounded packet | Live stage-ladder recommendation state above the bounded browser recognition shell. | | Provider-backed recognition service contract | Implemented now | first-party recognition client surfaces | First-party normalized recognition session boundary is real. | | Multilingual recognition guidance shell | Deep-source grounded retained | `qbr` retained remainder | Kept deferred until a font-safe and provenance-safe shipping path is explicit. | -| Multi-face correction/explanation shell | Deep-source grounded retained | `qbr`, `rubix-cube-solver` retained remainder | Not yet widened into the live shell. | +| Multi-face correction/explanation shell | Deep-source grounded retained | `rubix-cube-solver` with `qbr` capture-state adjunct value | Shared retained shell is source-backed, but keep it above the landed `Phase 6R-O`, `Phase 6R-P`, and `Phase 6R-Q` seams and keep both bundled-asset guards explicit. | ### 3. Training, coaching, and progression cockpit @@ -214,9 +215,10 @@ repo. | Feature | Status | Primary authority | Notes | |---|---|---|---| -| Session continuity and workspace recall fragments | Implemented now | first-party runtime/training surfaces | Real but not yet a full memory federation. | +| Session continuity and workspace recall fragments | Implemented now | first-party runtime/training surfaces | Real substrate fragments under the canonical `Memory Lanes`; not yet a full lane implementation. | | Provenance-aware training/replay/publication state | Implemented now | first-party contract/provenance surfaces | Existing product truth. | | Layered memory federation | Deep-source grounded retained | memory doctrine | Governing taxonomy exists; lane widening remains future work. | +| Continuity Lattice context assembly | Deep-source grounded retained | continuity-lattice doctrine | One federated substrate with two preset-backed, overrideable profiles: `Max-Retention Mode` and `Economic-Retention Mode`. This is not a second memory system and does not weaken lane authority or optional-assistive override control. | | User-authored note lane | Deep-source grounded retained | memory doctrine, owner unresolved | Do not describe as shipped. | | Compact memory views / reducers | Deep-source grounded retained | optional-assistive + memory doctrine | Must remain derived and optional. | diff --git a/docs/v6_5_deep_manual_pack/HyperTwist/PROVENANCE_AND_TRUST_MODEL.md b/docs/v6_5_deep_manual_pack/HyperTwist/PROVENANCE_AND_TRUST_MODEL.md index 64bbbad..7d5d2f9 100644 --- a/docs/v6_5_deep_manual_pack/HyperTwist/PROVENANCE_AND_TRUST_MODEL.md +++ b/docs/v6_5_deep_manual_pack/HyperTwist/PROVENANCE_AND_TRUST_MODEL.md @@ -84,6 +84,7 @@ The governing authorities are: - `C:\HyperTwist\docs\ops\HYPERTWIST_OPTIONAL_ASSISTIVE_FEATURE_DEACTIVATION_AND_REMOVABILITY_DOCTRINE_2026-05-21.md` - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` Companion extraction surface: @@ -100,6 +101,15 @@ and requires provenance to survive: - knowledge promotion - derived compact views +Important continuity-profile reading: + +- `Max-Retention Mode` and `Economic-Retention Mode` are transmission and + context-assembly profiles only +- they do not change provenance rank, custody visibility, memory ownership, or + truth arbitration +- confidence does not outrank authority +- a high-confidence derived record must not outrank a raw authoritative record + ## Redaction and publication Publication surfaces should preserve enough provenance to remain auditable diff --git a/docs/v6_5_deep_manual_pack/HyperTwist/ROADMAP.md b/docs/v6_5_deep_manual_pack/HyperTwist/ROADMAP.md index 9df5e54..2ebdfa5 100644 --- a/docs/v6_5_deep_manual_pack/HyperTwist/ROADMAP.md +++ b/docs/v6_5_deep_manual_pack/HyperTwist/ROADMAP.md @@ -9,6 +9,8 @@ Canonical discovery surfaces for roadmap interpretation: current landed runtime anchors - `C:\HyperTwist\docs\ops\HYPERTWIST_MEMORY_LANE_AUTHORITY_AND_PHASE_IMPLEMENTATION_DOCTRINE_2026-05-21.md` for memory-lane sequencing +- `C:\HyperTwist\docs\ops\HYPERTWIST_CONTINUITY_LATTICE_AND_CONTEXT_ASSEMBLY_PROFILE_DOCTRINE_2026-05-23.md` + for continuity substrate and context-assembly profile sequencing - `C:\HyperTwist\docs\ops\HYPERTWIST_PROVIDER_NEUTRALITY_AND_BYOK_DOCTRINE_2026-05-21.md` for provider/BYOK sequencing - `C:\HyperTwist\docs\ops\HYPERTWIST_SKILLIZATION_AND_COMMAND_SURFACE_DOCTRINE_2026-05-21.md` @@ -126,8 +128,13 @@ Canonical discovery surfaces for roadmap interpretation: packet is now landed in current code - the bounded permissive `Phase 6R-Q` `vivaansinghvi07/rubix-cube-solver` solve explanation/recommendation packet is now landed in current code -- the current next bounded move is a source-backed retained classic-cube recognition multi-face - correction/explanation preparation/control pass, while keeping both the `twistysim.min.js` and +- the continuity-profile sync now names `Continuity Lattice` as the substrate only, keeps + `Memory Lanes` canonical, and treats `Max-Retention Mode` and `Economic-Retention Mode` as + preset-backed overrideable context-assembly profiles rather than a second memory system +- the generic source-backed `Phase 6R-R` shared classic-cube recognition multi-face + correction/explanation control pass is now consumed +- the current next bounded move is the bounded permissive `Phase 6R-R` classic-cube recognition + multi-face correction/explanation shell packet, while keeping both the `twistysim.min.js` and `arial-unicode-ms.ttf` redistribution guards explicit - the repo-row implementation queue is now live from that `Phase 6R-A` entry point rather than waiting on another first-party packet @@ -190,8 +197,8 @@ Current routing truth: - active non-live implementation-board rows: `34` - retained benchmark, oracle, or clean-room-later rows outside the active implementation board: `9` -The next bounded move is a source-backed retained classic-cube recognition multi-face -correction/explanation preparation/control pass shared across `kkoomen/qbr` and +The next bounded move is the bounded permissive `Phase 6R-R` classic-cube recognition +multi-face correction/explanation shell packet shared across `kkoomen/qbr` and `vivaansinghvi07/rubix-cube-solver`. Queue interpretation after that packet: @@ -216,7 +223,8 @@ Queue interpretation after that packet: - classic-cube webcam UI shell - still deferred: - bundled-font redistribution and multilingual solve shell - - multi-face correction / explanation ownership + - no standalone correction packet remains; the surviving adjacent value is + the shared `Phase 6R-R` capture-state adjunct - `roice3/Magic120Cell` remains a partially landed row rather than a closed row: - landed: - dedicated `120-cell` family runtime profile @@ -256,6 +264,10 @@ Queue interpretation after that packet: - `kkoomen/qbr` - do not copy or redistribute `src/assets/arial-unicode-ms.ttf` without separate confirmation or replacement in any future widening that would ship that asset + - `Phase 6R-R` + - keep `vivaansinghvi07/rubix-cube-solver` as the retained correction/explanation-semantic + owner and preserve `kkoomen/qbr` as the retained capture-state and re-scan adjunct without + widening into either bundled asset family - `roice3/MagicTile` remains a partially landed row rather than a closed row: - landed: - tiling topology and geometry-family contract @@ -322,8 +334,8 @@ Queue interpretation after that packet: - the next queue shape is now: - retained classic-cube recognition multi-face correction/explanation shell remainder - the next bounded packet should stay narrow: - - multi-face correction/explanation preparation/control before widening into bundled - `twistysim.min.js` redistribution, generalized solver backend ownership, or any `qbr` + - first bounded `Phase 6R-R` correction-state and correction-explanation shell widening before + bundled `twistysim.min.js` redistribution, generalized solver backend ownership, or any `qbr` multilingual/font-shipping path - keep the speech-input / voice sidecar legal sequencing guard visible: - keep code-license judgments separate from model, voice, and payload-license review